A NOMALY DETECTION OF ENERGY CONSUMPTION IN

BUILDINGS : A REVIEW, CURRENT TRENDS AND NEW

PERSPECTIVES

A P REPRINT

Yassine Himeur1∗, Khalida Ghanem2 , Abdullah Alsalemi1 , Faycal Bensaali1

arXiv:2010.04560v5 [cs.CY] 12 Feb 2021

1
Department of Electrical Engineering
Qatar University, Doha, Qatar
yassine.himeur@qu.edu.qa;a.alsalemi@qu.edu.qa;f.bensaali@qu.edu.qa
2
Division Telecom, Center for Development of Advanced Technologies (CDTA),
Algiers Algeria
kghanem@cdta.dz

Abbes Amira
Institute of Artificial Intelligence
De Montfort University, Leicester, United Kingdom
abbes.amira@dmu.ac.uk

February 15, 2021

A BSTRACT

Enormous amounts of data are being produced everyday by sub-meters and smart sensors installed
in residential buildings. If leveraged properly, that data could assist end-users, energy producers
and utility companies in detecting anomalous power consumption and understanding the causes of
each anomaly. Therefore, anomaly detection could stop a minor problem becoming overwhelming.
Moreover, it will aid in better decision-making to reduce wasted energy and promote sustainable and
energy efficient behavior. In this regard, this paper is an in-depth review of existing anomaly detection
frameworks for building energy consumption based on artificial intelligence. Specifically, an extensive
survey is presented, in which a comprehensive taxonomy is introduced to classify existing algorithms
based on different modules and parameters adopted, such as machine learning algorithms, feature
extraction approaches, anomaly detection levels, computing platforms and application scenarios. To
the best of the authors’ knowledge, this is the first review article that discusses anomaly detection
in building energy consumption. Moving forward, important findings along with domain-specific
problems, difficulties and challenges that remain unresolved are thoroughly discussed, including the
absence of: (i) precise definitions of anomalous power consumption, (ii) annotated datasets, (iii)
unified metrics to assess the performance of existing solutions, (iv) platforms for reproducibility and
(v) privacy-preservation. Following, insights about current research trends are discussed to widen the
applications and effectiveness of the anomaly detection technology before deriving future directions
attracting significant attention. This article serves as a comprehensive reference to understand
the current technological progress in anomaly detection of energy consumption based on artificial
intelligence.

Keywords Energy consumption in buildings · anomaly detection · machine learning · deep abnormality detection ·
energy saving.

∗
Applied Energy Volume 287, 1 April 2021, 116601
 A PREPRINT - F EBRUARY 15, 2021

1 Introduction

Nomenclature MCU microcontroller unit

ANN artificial neural network MNN mutual k-nearest neighbor
AIN artificial immune network MLP multi-layer perceptron
ARIMA autoregressive integrated moving average MSCRED multi-scale convolutional recurrent encoder-
CBLOF cluster-based local outlier factor decoder
CNN convolutional neural network MSE multiview stacking ensemble
DAD deep abnormality detection NILM non-intrusive load monitoring
DAE deep autoencoder NTL non-technical loss
DBN deep belief network OCL one-class learning
DBSCAN density-based spatial clustering of OCSVM one-class support vector machine
applications with noise OCNN one-class neural network
DODDS distance-based outlier detection in OCCNN one-class convolutional neural network
data streams OCRF one-class random forest
DNN deep neural networks PCA principal component analysis
DRED Dutch residential energy dataset PGBOD parallel graph-based outlier detection
DRL deep reinforcement learning PIR passive-infrared
ELM extreme learning machines PCSiD power consumption simulated dataset
GAN generative adversarial networks QDA quadratic discriminant analysis
GBM gradient boosting machine QUD Qatar university dataset
GTB gradient tree boosting RBFNN radial basis function neural network
iForest isolated forest RBM restricted Boltzmann machine
IoT Internet of things RNN recurrent neural network
KNN K-nearest neighbors ROF resolution-based outlier factor
KNNG K-nearest neighbor graphs SCiForest isolation forest with split-selection criterion
LDA linear discriminant analysis semi-SVM semi-supervised support vector machine
LDCOF local density cluster-based outlier factor SLFN single-layer feed-forward neural network
LOESS locally estimated scatterplot smoothing SVM support vector machine
LOF local outlier factor SVR support vector regression
LSTM long short-term memory STTS short-term time-series
MDA multiple discriminant analysis VFD variance fractal dimension
(EM)3 Consumer Engagement towards Energy saving behavior by means of Exploiting Micro Moments
and Mobile recommendation systems

Climate change is an dangerous predicament affecting the world’s population. Almost 80% of the overall world energy
is produced by fossil fuels. In addition to find green energy sources, it is of utmost importance to diminish the total
energy consumption percentage [1]. A notable approach into achieving this objective is through informing end-users of
their power usage patterns. Accordingly, consumers can improve their behavior and change their consumption habits
with the aim of reducing wasted energy and contributing in the promotion of sustainable and green energy ecosystems
[2]. This is quite possible, especially if recommender systems are combined with anomaly detection modules. Therefore,
personalized and contextual recommendations will be generated and transmitted the end-user to assist them in adopting
a more sustainable energy use behavior [3, 4]. In this line, governments around the world have realized the importance
of energy efficiency and the major role that end-users can play to curtail the entire expenditure on energy [5].
On the other side, the building sector represents a major energy consumer across the world. Specifically, buildings are
responsible of more than 40% of the overall energy generated globally, which is converted to more than 30% of the
entire worldwide CO2 emission [6, 7]. As such, the reduction of power consumption in building environments could
absolutely support the urgently-needed diminutions in the world-wide power consumption and the related environmental
interests. Nevertheless, reducing power consumption in buildings is not straightforward and is a challenging task since
each building requires electrical energy to operate [8, 9]. Even though there is an increasing interest towards developing
zero-energy buildings, related ideas are only in their nascence and are just tested in limited regions of developed
countries. In this context, the potential option available currently is to promote energy awareness and optimize the
operation of appliances used in buildings, giving that the latter are rigorously built to consume the amount of energy
needed for their expected aims, i.e. preventing energy waste [10, 11]. According to recent studies, people could
spend up to 80–90 % of their time in indoor environments (and extensively some unexpected circumstances, such as
the COVID-19 pandemic), which can enormously impact their energy consumption levels, especially if they show
negligence and carelessness [12, 13].

2
 A PREPRINT - F EBRUARY 15, 2021

Efficient feedback could help in reducing energy consumption in buildings and lessening CO2 emissions. Accordingly,
offering updated information and personalized recommendations to end-users and building managers is the initial
stage towards setting innovative approaches to optimize energy usage [14, 15]. In addition, for effective power usage,
anomalous consumption behavior must be captured [16]. Therefore, via implementing energy monitoring systems and
benchmarking strategies, abnormal behavior and footprints can be mitigated. Consequently, smart anomaly detection
techniques for energy consumption should be formulated for identifying new forms of abnormal consumption behaviors
[17]. In buildings, an anomalous behavior of an electrical device or of the end-user could occur either because of a
faulty operation of a device, end-user negligence (e.g. cold loss in a room by keeping a window open while the air
conditioner is on or refrigerant leak in a fridge via maintaining the fridge door open), a theft attack, a non-technical loss,
etc. [18, 19]. An occurrence of anomalous behavior could lead to higher power consumption, longer operation-time
than its normal behavior/operation-time and/or could result in a permanent malfunction of the device [20].
It has been demonstrated in various research works that it should be possible to utilize artificial intelligence (AI)
for detecting anomalous energy consumption behaviors either generated by end-users, appliances’ failures, or other
potential causes [9, 21]. The AI community has made every possible effort during the past decade to detect abnormal
power consumption accurately and swiftly. However, it is also of significant importance to detect when an appliance is
not working appropriately and what are the reasons. Moreover, energy consumption events occurring during a day-off
may be genuine, or harder to deal with compared to recurring events, and thus an anomaly detection algorithm might
consider a recurring fault as “normal”. This makes anomaly detection in energy consumption very different form
other application scenarios, e.g. intrusion detection, healthcare anomaly detection, etc. [22]. This is because (i) the
other applications are drastically different as they have acute, serious consequences if the anomaly is not detected,
whereas household energy anomalies might cause extract costs and jumps of the energy bills every month, but are
unlikely to be life threatening; and (ii) detecting anomalous consumption should be followed by triggering a set of
tailored recommendations to help end-users adjust their energy consumption habits, replace faulty appliances, identify
cyber attackers on energy infrastructures and carry on legal procedures and take other measures related to end-users’
negligence (e.g. close the refrigerator door, close the doors and windows of the rooms while an air conditioner is
working, etc.) [23]. Such measures could be quite useful in different ways since they result in high energy cost savings,
and could further prevent different kind of disasters (e.g. a house fire).
Efficient energy saving systems based on anomaly detection schemes need to address various issues before reaching a
wider adoption. Among the challenges is how to design scalable and low cost solutions while maintaining decentral-
ization and security. Other contemporary issues include privacy preservation, consumer anonymity, and the real-time
implementation of anomaly detection based systems. A significant effort has been put in recent years to innovate
anomaly detection strategies, a large amount of projects and frameworks are ongoing, which have been described
in scientific journal articles, patents, reports and industrial white papers and produced principally by the academic
community and industrial partners. Moreover, various AI-based anomaly detection techniques have been the subject of
new energy saving solutions. However, we assert a systemic and comprehensive review conducted based on different
sources is still required to investigate the challenges, issues and future perspectives of the applicability of machine
learning for anomaly detection in energy consumption. In this context, this framework strives to fill that knowledge
gap via proposing, to the best of the authors’ knowledge, the first, extensive and timely survey of anomaly detection
of energy consumption in buildings. Explicitly, with the aim of laying the foundation for this effort, the following
contributions have been proposed:
• First, we present an overview of existing anomaly detection schemes in building energy consumption, in
which a comprehensive taxonomy is adopted to classify them into various categories based on the nature
of machine learning model used to identify the anomalies, feature extraction, detection level, computing
platform, application scenario and privacy preservation. In addition, we discuss various system architectures
and associated modules determining the technical properties of anomaly detection systems. A considerable
part of current knowledge on anomaly detection in energy consumption arises not just from conventional
academic sources (i.e. journal articles and conference proceedings), but also from industrial outputs, granted
patents, and whited papers. We focus in the first part of this framework on distilling valuable information
from the aforementioned sources in order to allow the readers comprehending the technical challenges of
energy consumption anomaly detection. More specifically, the advantages and limitations of every category is
discussed thoroughly along with its competence in different case scenarios.
• Second, we perform a critical analysis and describe by conducting an in-depth discussion of the presented
state-of-the-art. We explore current difficulties and limitations issues associated with the development and
implementation of the anomaly detection systems, in addition to their market barriers.
• Third, we describe current trends and identify new challenges concerning the enrichment of anomaly detection
schemes with new applications and functionalities that could impact positively the energy consumption in
buildings, among them considering additional sources of data (e.g. occupancy patterns, ambient conditions,

3
 A PREPRINT - F EBRUARY 15, 2021

etc.), combining other technologies (i.g. non-intrusive load monitoring (NILM)), collecting annotated datasets
and using unified assessment metrics.
• Finally, we derive a set of future research directions that require greater emphasis with regard to four aspects,
in order to (i) overcome the actual drawbacks of anomaly detection algorithms, (ii) improve the exploitation of
anomaly detection solutions for better energy saving ecosystems, (iii) improve the deployment of innovative
anomaly detection systems in real-world scenarios, and (iv) preserving the privacy of end-users.

The remainder of this paper is organized as follows. An overview of state-of-the-art anomaly detection techniques in
building energy consumption is presented in Section 2, where an exhaustive taxonomy is proposed with regards to
various aspects. Furthermore, their limitations and drawbacks are highlighted. Moving forward, critical analysis and
discussion are presented in Section 3 as a result of the conducted overview, in which difficulties, limitations and market
barriers are described. Following, Section 4 is divided into two parts, in which Section 4.1 is reserved to describing
open research challenges regarding novel applications and functionalities of anomaly detection methods. While, Section
4.2 provides a set of insightful perspectives and emerging concepts for advancing future anomaly detection systems.
Finally, Section 5 derives relevant concluding remarks.

2 Overview of anomaly detection methods

2.1 Overview

This section describes existing anomaly detection methods based on the nature of implemented AI algorithms used
to detect anomalies. Fig. 1 illustrates the proposed taxonomy of anomaly detection techniques in building energy
consumption with reference to different aspects.

2.1.1 Unsupervised detection (U)

It aims at detecting formerly unknown rare consumption observations or patterns without using any a priori knowledge
of these observations. Generally, this kind of detection assumes that the amount of anomaly patterns to the overall
consumption data is small, i.e. less than 20%. Since the abnormalities represent the outliers that are unknown to the
consumer at the training stage, detecting anomalous consumption is reduced to the modeling of normal consumption
behavior in the large majority of cases, in addition to the definition of specific measurements in this space with the
aim of classifying consumption observations as abnormal or normal. Unsupervised techniques are mainly built on
clustering, one-class learning and dimensionality reduction algorithms.

U1. Clustering: it is a machine learning scheme used to split power consumption data into various clusters and hence
helps in classifying them into normal or abnormal in unlabelled datasets (even with many dimensions). This anomaly
detection strategy has attracted a lot of interest in different research topics for its simplicity, such as intrusion detection
in networks [24], Internet of things (IoT) [25], sensor networks [26], suspicious behavior detection in video surveillance
[27], anomalous transaction detection in banking systems [28] and suspicious account detection in online social
networks [29]. In addition, clustering has the capability for learning and detecting anomalies from the consumption’s
time-series without explicit descriptions [30].
Aiming at distinguishing between actual anomalies and genuine changes due to seasonal variations, the authors in
[31] propose a two-step clustering algorithm. In the first step, an anomaly score pertaining to each user is periodically
evaluated by just considering his energy consumption and its variations in the past, whilst this score is adjusted in the
second step by taking into account the energy consumption data in the neighborhood. In [32], the concept of “collective
anomaly” is introduced, instead of the events that refer to an anomaly, to depict itemsets of events, which, depending on
their patterns of appearance, might be anomalous. To achieve this, the frequent itemset mining and categorical clustering
with clustering silhouette thresholding approaches were applied on smart meters data streams. In [33] an integrated
scalable framework which combines clustering and classification techniques with parallel computing capabilities is
adopted, by superimposing a k-means model for separating anomalous and normal events in highly coherent clusters.
Moving forward, authors in paper [34] opt for a time-series to investigate the anomaly detection in temporal domain,
subsequently to categorizing the anomalies into amplitude and shape related-ones. A unified framework is introduced
to detect both type of anomalies, by employing fuzzy C-means clustering algorithm to unveil the available normal
structures within the subsequences, along with a reconstruction criterion implemented to measure the dissimilarity
of each subsequence to the different cluster centers. In [35], power data are processed through the mutual k-nearest
neighbor (MNN) and k-means clustering algorithms to reduce the number of measurement samples, the consumption
patterns are then analyzed to detect abnormal behaviors and malicious customers. Finally, entropy-based methods

4
 A PREPRINT - F EBRUARY 15, 2021

Figure 1: Taxonomy of the anomaly detection schemes in energy consumption based on different aspects: i) AI
algorithms, ii) application scenarios, iii) detection levels and iv) computing platforms.

for anomaly detection represent another clustering category, in which a little effort has been devoted to thoroughly
comprehend the detection force of using entropy-based analysis, such as [36, 37].

U2. One-class classification: also named one-class learning (OCL) relies on considering initial power consumption
patterns to be parts of two groups, positive (normal) and negative (abnormal), then it attempts to design classification
algorithms while the negative group can be either absent, poorly sampled or unclear [38]. Accordingly, OCL is
a challenging classification problem that is harder to solve than conventional classification problems, which try to
discriminate between data from two or more categories using training consumption data that pertain to all the groups
[39].
Different schemes have been proposed in the literature to detect anomalous consumption footprints based on OCL. In
[40], one-class support vector machine (OCSVM) is introduced to identify the smallest hypersphere encompassing all
the power observations. In [41], a kernel based one-class neural network (OCNN) is proposed to detect abnormal power
consumption. It merges the capability of deep neural networks (DNN) to derive progressive rich representations of

5
 A PREPRINT - F EBRUARY 15, 2021

power signals with OCL, building a tight envelope surrounding normal power consumption patterns. In [42, 43], two
different approaches of one-class convolutional neural networks (OCCNN) are proposed. They share the same idea of
using a zero centered Gaussian noise in the latent space as the pseudo-negative class and training the model based on
the cross-entropy loss to learn an accurate representation along with the decision boundary for the considered class.
Also, one-class random forest (OCRF) is proposed to identify abnormal consumption when labeled data are absent
[44, 45], it is based on utilizing classifier ensemble randomization fundamentals [46].

U3. Dimensionality reduction: in different machine learning applications, dimensionality reduction could be used as
a classification approach with a low computational cost as it can removes irrelevant power patterns and redundancy [47].
Various techniques are explored to classify power data as normal or abnormal, such as principal component analysis
(PCA), linear discriminant analysis (LDA) [48], quadratic discriminant analysis (QDA) [49] and multiple discriminant
analysis (MDA) [50].
Despite the fact that PCA has been proposed mainly to reduce the dimensions of the original data while preserving the
relationships between the data as much as possible, it has been also used as a classifier. For example, in the anomaly
detection problem that is considered as a two-class classification issue, the PCA classifier estimates the principal
components of both normal and abnormal classes. Following, the classifier is designed with reference to the projection
of the energy patterns within the subspaces spanned by these principal components for either normal or abnormal class
[51, 52]. Moreover, PCA could also be applied for the case of multi-class anomaly detection, as it is the case with
the micro-moment based anomaly detection approach described in [19]. Accordingly, the normal energy usage class
has been split into three new classes while abnormal energy consumption class has been divided into two new classes.
Overall, the anomaly detection problem has become a classification issue of 5 different classes. All in all, PCA is
appropriate for the case in which energy observations of different categories are distributed in different spaces and
directions.
In [53], the Karhunen-Loeve transform-based PCA is used to detect anomalous power consumption. It relies on
estimating principal components of every consumption category and then creates a classifier via projecting power
patterns on the subsets distributed by those principal components related to the two main categories (i.e. normal
and abnormal). In [54], LDA is used to classify power consumption patterns by discriminating between separated
sub-categories and design a model to automatically labeling power consumption patterns with reference to their
corresponding categories. This has been accomplished via the use of discriminant weights to separate the hyperplanes
generated by the LDA statistical learning. In [55, 56], QDA that is a variant of LDA is deployed to enable a non-linear
separation of power consumption patterns pertaining to both normal and abnormal ensembles. Finally, MDA is mainly
used to build discriminant axes (functions) from linear combinations of the initial power consumption data. Every axis
is designed to maximize the difference between normal and abnormal categories while considering them uncorrelated
[56, 57].

2.1.2 Supervised detection (S)

Supervised anomaly detection in energy consumption necessitates training the machine learning classifiers (binary
or multi-class) using annotated datasets, where both normal and abnormal power consumption is labeled. Although
supervised anomaly detection can achieve high-accuracy identification results as demonstrated in academic frameworks,
its adoption in the real world is still limited compared to unsupervised methods, due to the absence of power consumption
annotated datasets. Fig. 2 illustrates the main steps to conduct a supervised anomaly detection approach.

S1. Neural networks: refer to using deep learning or conventional artificial neural networks (ANN) to detect normal
and abnormal consumption patterns. Currently, deep abnormality learning (DAD) has been used in various research
topics, such as detecting fraudulent health-care transactions [41], identifying abnormalities in video streaming [58] and
detecting credit card frauds [59]. However, the performance of a deep learning based solution could be sub-optimal in
some cases owing to the imbalance property of power consumption datasets (i.e. power consumption patterns are not
uniformly distributed over the normal and abnormal categories).
In [60, 61], the autoencoder and long short-term memory (LSTM) neural networks are merged to identify abnormalities
in unbalanced and temporally correlated power consumption datasets. Similarly, in [62], the authors detect anomalies in
time-series power footprints using a variational recurrent autoencoder. Moving forward, Yuan et Jia [63] use stacked
sparse autoencoder for extracting high-level representations from large-scale power consumption datasets gleaned using
and IoT-based metering network. Next, they utilize softmax in the classification stage to capture the consumption
anomalies before sending notifications and alerts to end-users using web applications. Similarly, in [64] the autoencoder
and micro-moment analysis are used to detect abnormal energy usage.

6
 A PREPRINT - F EBRUARY 15, 2021

Figure 2: The main steps used to perform a supervised anomaly detection of energy consumption in buildings.

On the other side, convolutional neural networks (CNN) have demonstrated its effectiveness in different research
applications, and it has superior performance in comparison with artificial neural network (ANN) algorithms for
detecting abnormalities in time-series data [65]. In [66], the author opted for combining CNN and random forest to
track energy consumption anomalies due to energy theft attacks and thereby helping energy providers to remedy the
issues related to irregular energy usage and inefficient electricity inspection. Similarly, Zheng et al. [67] propose a
CNN-based solution, which helps mainly in identifying the non-periodicity of energy theft and periodicity of normal
energy consumption using 2D representations of power consumption signals. Using the same idea, a CNN is developed
in [68] via representing time-series time/frequency energy consumption signals in 2D space and then learning anomaly
features using convolution. Moving forward, in [69], multi-scale convolutional recurrent encoder-decoder (MSCRED)
is deployed to analyze multivariate time-series observations and detect abnormalities. In [70], a restricted Boltzmann
machine (RBM) along with a deep belief network (DBN) are merged to construct a DNN-based abnormality detection
framework. Explicitly, a dimensionality reduction task is performed at the two first RBM layers before being fed into a
fine tuning layer including a classifier to separate anomalies from normal data.
Furthermore, looking for innovative deep learning solutions to deal with the unbalanced property of anomaly detection
datasets, generative adversarial networks (GAN) are employed. It can model complex and high-dimensional data of
different types, including images [71], time-series [72, 73] and cyber security [74]. Unfortunately, its utilization to
detect anomalous power consumption in buildings is still very limited [75].
Recurrent neural network (RNN) is very competent in analyzing time-series data and enables to exhibiting temporal
dynamic behaviors [76]. It has been used to predict the anomalies occurring during energy usage and distinguish them
from deviations emerging from seasonality, weather and holiday dependencies [77, 61]. For instance, in [78], an RNN
based anomaly detection system is designed, which can remove seasonality and trends from power consumption patterns,
resulting in a better capture of the real abnormalities. In [79], the authors concentrate on elaborating an abnormality
detection scheme having the ability to face the concept drift, due to family structure changes (e.g. a household turned
to a second family residence). To that end, an LSTM based RNN model is developed to profiling and forecasting
end-users’ consumption behaviors using their recent/past consumption data. In [80], abnormal days illustrate suspicious
consumption rates are identified using a hybrid learning model based on RNN and K-means. Similarly in [81], a hybrid
model using RNN and quantile regression is introduced to predict and detect anomalous power consumption.

7
 A PREPRINT - F EBRUARY 15, 2021

Figure 3: Flowchart of a supervised anomaly detection scheme based on DNN used to detect two different anomalies,
i.e. excessive consumption and consumption while outside.

In order to provide the reader with more details on the use of deep learning for anomaly detection in energy consumption,
Fig. 3 illustrates a flowchart of a supervised anomaly detection scheme proposed in the (EM)3 project, which is
performed using a DNN model [19]. In this framework, power consumption data of various appliances and occupancy
patterns are gleaned using sub-meters and smart sensors. Next, collected data are labeled using a micro-moment
paradigm, in which consumption footprints are divided into five consumption categories. Following, a DNN model is
designed and trained using the labeled dataset before testing it on new recorded, unlabeled data in the test stage.

On the other hand, using ANN for anomaly detection in energy consumption is mainly supported by its capability to
learn and generalize from past consumption data to identify normal and abnormal behavior [82]. In addition, ANN could
help in solving the anomaly detection issue when recorded data is noisy due to various reasons, e.g. noise generated
during data transmission or from electrical appliances connected to the smart grid [83]. In [84], the identification of
power consumption anomaly is handled by resorting to a multi-stage ANN-based solution. This latter incorporates a
discrete wavelet transform to obtain the required features, a variance fractal dimension (VFD) operation applied on
those features, an ANN scheme which exploits the VFD output to perform the training, and finally a threshold-based
detection of the anomalous power consumption pattern. The work in [85] proposes a residential framework comprising
a dual hybrid one-step-ahead load predictor and a rule-engine-based energy consumption abnormality detector. In order
to attain a high anomaly detection precision in linear and nonlinear regression, the predictor merges the benefits of
ANN and autoregressive integrated moving average (ARIMA) model.
Moreover, the consumption anomalies are tracked through the use of multi-layer perceptron (MLP) and classification
techniques in [86]. Similarly in [87], with the aim of predicting malicious behavior in unbalanced data, an MLP-based
solution is efficiently tested on two different datasets to carry out a flow-based control which preserves the end-users’
privacy. In the same direction, the continuous and fine-grained monitoring of energy consumption in industrial buildings
is discussed in [88] in order to preserve reliable operation. Explicitly, an MLP-based anomaly detection scheme is
targeted via detecting sensor data abnormalities in a pharma packaging system. Moreover, intrusion detection that can
be applied in energy theft tracking, is investigated in [89] by combining artificial immune network (AIN) and cosine
radial basis function neural network (RBFNN), wherein firstly multiple-granularities version of the former is supported
to reveal the candidate hidden neurons, and subsequently, the latter is trained based on gradient descent learning process.
In addition, different power consumption anomaly detection frameworks are introduced based on extreme learning
machines (ELM) [90, 91]. Specifically, ELM is built upon a single-layer feed-forward neural network (SLFN) for
classifying the normal and abnormal classes [92].

8
 A PREPRINT - F EBRUARY 15, 2021

S2. Regression: refers to identifying the relationship between two or more power variable classes in order to produce an
ensemble of model parameters to predict the generation of abnormal power observations. In this context, the production
of anomalous power consumption patterns can be predicted based on other collected abnormal footprints. Various
regression models have been introduced in the literature to identify abnormalities in building energy consumption,
including linear regression, support vector regression (SVR), auto-regressive models, regression trees and regression
fitting [93]. The authors in [94] propose to adopt linear regression-based approaches to determine the anomalous periods
for individual premises, and clear them from the premise data, such that to provide precise assessments of energy
consumption patterns. In the same direction, a model to find abnormal energy consumption patterns is designed in [95]
by analyzing the smart meters temporal data streams. Specifically, to perform the prediction and map the non-linearity
of data, support vector regression with radial basis function is retained and correspondingly evaluates the disparity
between the actual and the expected energy consumption.
Because of the large quantity of stored smart meter data, anomaly detection with such information brought the big data
issue into focus, particularly with the scarcity of adequate and efficient real time anomaly detection systems capable of
handling this huge amount of data. In order to remedy this and facilitate energy-related decision-makings, the studies
in [96, 97] depict a scalable architecture merging an autoregressive prediction-based detection method, with a new
lambda scheme to iteratively upgrade the model along with real time anomaly detection. The work in [98] targets the
reduction of anomalous consumption by presenting a new scheme which enabled the identification of anomalous power
consumption within large sets of data. It follows a two-stage processing, namely prediction and then anomaly detection,
where, by the aid of a hybrid neural network ARIMA model of daily consumption, daily real-time consumption is
first predicted in the former step, whereas a two-sigma rule was adopted to localize the anomalies via the evaluation
of the mismatch between real and predicted consumption. The framework in [65] address the anomaly recognition in
streaming large scale data, which is a typical occurrence scenario in deployed sensors. In this scope, both statistical (i.e.
ARIMA) and CNN based approaches were integrated in a residual way, such that the fusion was shown to compensate
the weaknesses of each of them and consolidate their strengths. In [88], a data-driven approach was pursued since
no cyclicity pattern was noted on the observed data. From comparing three different regressors (i.e. regression tree,
random forest, and MLP) in the prediction phase, the authors highlighted the advantages of the regression trees and
random forests residing in the training time efficiency and model replicability ease.

S3. Probabilistic models: are among the most important machine learning tools, they have been instituted as an
effective idiom for describing the real-world problems of anomaly detection in energy consumption using randomly
generated variables, such as building models represented by probabilistic relationships [99, 100]. The anomaly profiles
of time-series patterns are identified using Bayesian maximum likelihood models for clean data [101] and noisy data
[100], while Bayesian network models are implemented to detect abnormalities categorical and mixed based power
consumption data in [102, 103]. In [104, 105], statistical algorithms are deployed to identify the anomalies via the
identification of extremes based on the standard deviation, while in [104], the authors use both statistical models and
clustering schemes to detect power consumption anomalies. In [106, 107], naive Bayes algorithms are proposed to
detect the abnormalities generated by electricity theft attacks. Similarly in [108], Janakiram et al. deploy a belief
Bayesian network to capture the conditional dependencies between data and then identify the anomalies. In [109], a
statistical prediction approach based on a generalized additive model is introduced to timely detect abnormal energy
consumption behavior.

S4. Traditional classification: stands for models that rely on detecting to which power consumption category (sub-
population) a new power consumption sample pertains, with reference to a training ensemble of consumption footprints
that have labels of both normal and anomalous consumptions. K-nearest neighbors (KNN), support vector machine
(SVM), decision tree and logistic regression are the well-known conventional classification algorithms, they have been
widely deployed in the state-of-the art of the energy-based applications or other research topics.
In [104, 53], KNN based heuristics are proposed to detect abnormal power consumption, while in [86], the authors
investigate the performance of KNN against other machine learning classifiers to identify abnormal power observations.
In [110, 111], SVM is deployed to detect abnormalities due to energy theft attacks. In the same direction, in [112], a
genetic SVM model is proposed to detect abnormal consumption data and suspicious customers, in which a genetic
algorithm is combined with SVM. While in [113], Zhang et al. fuse SVM and particle swarm optimization for detecting
abnormal power consumption in advanced metering infrastructures. On the other side, in [114], a decision tree based
solution is introduced to learn energy consumption anomalies triggered by fraud energy usage. Similarly in [115], an
improved decision tree model is developed to detect anomalous consumption data using densities of the anomaly and
normal classes. Moving forward, in [88], a decision tree regressor is presented to detect abnormal power consumption
using sensor data, while in [86], the anomalies are detected using logistic regression.

9
 A PREPRINT - F EBRUARY 15, 2021

2.1.3 Ensemble methods (E)

As it is demonstrated in various frameworks [19, 116], none of the anomaly detection schemes could perfectly identify
all abnormalities through low-dimensional subspaces because of the complexity of power consumption data and other
factors influencing power usage over hourly, daily, weekly, monthly or yearly scales. Accordingly, the use of ensemble
learning can solve some related issues, where the initial set of power observations is split to multiple subsets and
various models are applied simultaneously on these subsets to derive the potential abnormalities. Following, anomaly
identification scores are either summarized or the most suitable one is selected to produce final score.

E1. Boosting: it is a set of meta-algorithms used to principally reduce bias and variance of unsupervised learning, in
which weak classifiers (learners) are converted into strong ones. Generally, they are structured in a sequential form. A
weak classifier refers to the case where a slight correlation can be achieved with the true classification [117]. Different
boosting schemes are proposed in literature to detect anomalies, among them bootstrap, gradient boosting machine
(GBM) and gradient tree boosting (GTB).
In [118], Zhang et al. use a bootstrap strategy to conduct an unlabeled learning process for detecting anomalies in energy
data in multi-feature data. In [119], a GBM based anomaly detection is introduced to model power usage of commercial
buildings. In the same manner, in [120], a grid search is deployed to capture the best parameter configuration of a
GBM based anomaly detection. While in [121], the authors predict energy frauds though the identification of power
consumption anomalies using a GBM based scheme. In [122], a GTB based anomaly detection is investigated along
with other data mining techniques using power consumption pricing data.

E2. Bagging: also called bootstrap-aggregating, it is a set meta-algorithms developed for improving the accuracy and
stability of several weak classifiers. Bagging differs from boosting by the fact that the weak learners are structured in a
parallel form [123]. Moreover, distinct detection schemes can be applied on each sub-ensemble before aggregating
their results as demonstrated in [124]. Random forests, bootstrap aggregation and their variations are the well-known
bagging based ensemble learning methods used for anomaly detection. For example, in [125], Araya et al. propose a
bootstrap aggregation based abnormality detection scheme, which helps in conducting an ensemble learning to identify
energy consumption anomalies. In [126], an isolation forest with split-selection criterion (SCiForest) algorithm is
introduced to check if the end-user’s electricity consumption is anomalous or normal. In [66], non-technical losses
(NTLs) occurring in the energy networks are detected using a random forest scheme. This is mainly conducted through
sensing anomalous power consumption and learning consumption differences for different periods (i.e. hours and days).
In [127], a random forest classifier is deployed to detect anomalies while respecting the performance measure related
to the accuracy and false alarm rates. In [128], a multiview stacking ensemble (MSE) technique is proposed to learn
energy consumption anomalies collected using different IoT sensors in industrial environments. In [116], an anomaly
detection scheme based on feature bagging is introduced. It relies on training several classifiers on different feature
sub-ensembles extracted from a main high-dimensional feature set and therefore combining the classifiers’ results into a
unique decision. In [129], after deriving various feature sub-ensembles randomly from the initial feature, anomalies are
identified and the performance is estimated in each sub-ensemble before fusing them to come out with the final output.

2.1.4 Feature extraction (F)

This part mainly discusses how feature extraction scheme can help to boost the performance of anomaly detection
methods via: (i) representing the power consumption observations in novel spaces (e.g. high-dimensional spaces); (ii)
utilizing appropriate measures and functions (e.g. distance, density) to discriminate between normal and abnormal
consumption; and (iii) representing the consumption flowchart using new representation structures (e.g. graph-based
representation) [130].

F1. Distance-based: refers to detecting abnormal consumption patterns by judging each pattern based on its distance
to its neighboring samples. Explicitly, normal consumption observations generally possess a dense neighborhood
while anomalous consumption footprints are far form their neighboring points (i.e. show a sparse structure). Various
frameworks have been proposed to resolve the issue of distance-based anomaly detection for energy consumption,
where unsupervised learning methods are usually adopted without having any distributive presumptions on recorded
consumption data. In this regard, in [131], a distance-based anomaly detection is proposed via analyzing the theoretical
properties of the nearest neighbors of each power observation. Explicitly, anomalous patterns are then detected with
reference to a global quantity named distance-to-measure. Also in [132], power anomalies in smart grid are detected
using a multi-feature fusion that is based on Euclidean distance and a fuzzy classification approach. In [133], the
authors use a cosine similarity approach to estimate similarity distance between power consumption observations and

10
 A PREPRINT - F EBRUARY 15, 2021

detect suspicious patterns. Following, they sort the resulted cosine distance data for identifying abnormal consumption
behavior based on a threshold.
Moreover, in [134], various methods are proposed to resolve the distance-based outlier detection in data streams
(DODDS) issue and their performance is compared when detecting anomalies without having any distributional
assumptions on power consumption observations. In a similar way, in [135], Huo et al. develop an distance-based
abnormality detection method, in which a time-space trade-off strategy has been deployed for reducing the computational
cost. While in [136], a resolution-based outlier factor (ROF) method is proposed to detect anomalies in large-scale
datasets. It mainly focuses on analyzing the distances of both local and global features to effectively detect anomalous
data. In [137], the energy consumption anomaly detection process is performed using an isolated forest (iForest) model.
The latter has been proposed by Liu et al. as a competitive method to ROF and local outlier factor (LOF) algorithms
[138, 46].

F2. Time-series analysis: because power consumption data are considered time-series footprints, it is logical that
many studies have focused on formulating the anomaly detection issue such as to find anomalous observations based
on standard signal analysis [128]. Specifically, this kind of anomaly detection relies on detecting unexpected spikes,
level shifts, drops and irregular signal forms. For example, in [139], seasonal trend decomposition using locally
estimated scatterplot smoothing (LOESS) is proposed to detect anomalous consumption points, in which a seasonal-
trend decomposition scheme based on LOESS is introduced. It helps in splitting the power consumption time series
samples into three components defined as seasonal, trend, and residue [140].
On the other side, it is worth noting that most of the anomaly detection schemes pertaining to this class are based
on short-term time-series (STTS) analysis. In this line, a log analysis of power consumption time-series patterns is
conducted in [141] to detect real-time anomalies in early warning systems. Similarly, [142], a feature extraction based
abnormality detection scheme is proposed using canonical correlation. It can help in detecting the anomalies in different
kinds of buildings, such as households, work spaces and industrial zones. In [143], abnormalities occurring in smart
meters data are identified using time-series analysis, in which Cook’s distance is deployed over a thresholding process
to decide whether an observation is normal or abnormal. In the same vein, in [144], a hierarchical feature extraction
method is proposed in order to capture energy consumption anomalies in time-series consumption data due to electricity
stealing. While in [145], to identify the abnormal consumption behavior, the authors analyze different STTS features
that could offer valuable details about deviations from a typical behavior.
On the flip side, other techniques use rule-based algorithms to analyze time-series data and detect anomalous power
consumption [146, 147]. For example, in [148], Yen et al. introduce a rule-based approach to analyze the phase
voltages and then decide which are the anomalous patterns using an ensemble of rules. In the same direction, in [149], a
rule-based algorithm is combined with a linear programming approach to detect anomalous electricity consumption
and hence identify the locations of potential energy theft attacks and/or faulty meters. In [150, 151], the detection
of anomalous power consumption is performed using a rule-based algorithm, which is elaborated based on machine
learning methods and the knowledge of energy saving experts. An ensemble of energy saving parameters is then
introduced to track abnormalities. While in [152], a rule-based algorithm is combined with an improved nearest
neighbor clustering approach to identify potential abnormal power consumption behaviors. In [19], a micro-moment
based algorithm is proposed to detect two kinds of power consumption anomalies, which are due to (i) excessive power
consumption, and (ii) consumption while the end-users are outside. The latter is responsible of wasting a large amount
of energy for a set of appliances, such as the air conditioner, heating system, fan, light lamp and desktop/laptop.

F3. Density-based: refers to anomaly detection methods that investigate the density of each power consumption
pattern and those of its neighborhood. Moving forward, a power observation is considered as anomalous if it has a
lower density compared to its neighbors [153]. Various techniques have been proposed in this regard; among them LOF
that attempts to derive a peripheral observation by using density of its surrounding space [154]; cluster-based local
outlier factor (CBLOF) that relies on detecting the anomalies using the size of its power consumption clusters, and the
density between each power observation and its closest cluster [155]; local density cluster-based outlier factor (LDCOF)
that represents an improved version of CBLOF, in which it applies a local density concept when allocating anomaly
scores[156]. In this context, in [157], a density-based spatial clustering of applications with noise (DBSCAN) approach
is introduced to detect anomalous power consumption in a wind farm environment. Overall, density-based anomaly
detection has been widely investigated in other fields, such as activity monitoring [158], machine fault detection [159],
financial and banking systems [160], etc., their application to detect abnormal energy usage has not been very successful
since other kinds of anomalies exist. Specifically, density based schemes could only identify energy consumption
outliers based on analyzing energy consumption levels without the possibility to detect other abnormalities, e.g. energy
consumption of some appliances (e.g. television, air conditioner, lamp, fan, etc.) while the end-user is absent.

11
 A PREPRINT - F EBRUARY 15, 2021

F4. Graph-based: before applying graph-based methods to detect power consumption abnormalities, consumption
data should be converted into a graph-based structure. Because there are no common standards to model this kind
of data, researchers use various schemes to design such a representation. For instance, the authors in [161, 162],
consider the house, power generator, electric network, rooms, and appliances as nodes; and edges stand for the existing
connection between a specific room and the operation of an appliance. Following, abnormalities resulting in a structural
change of the graph topology are detected, while a graph-based abnormality is defined as an unforeseen deviation to a
normative pattern.
Different graph-based abnormality detection (GBAD) algorithms have been proposed [163], where abnormal observa-
tions of structural data are identified in the information representing entities, actions and relationships. In [164], the
authors propose a graph-based method to discover contextual anomalies in sequential data. Explicitly, the nodes of the
graph are clustered into different categories, where each class includes only similar nodes. Following, anomalies are
detected via checking if adjacent observations pertain to the same class or not. Similarly, in [165], a parallel graph-based
outlier detection (PGBOD) technique is introduced for identifying power abnormalities, in which data are processed in
parallel before extracting abnormal patterns.

2.1.5 Hybrid learning (H)

Annotating normal power consumption is much easier than labeling anomalous patterns, consequently, hybrid or
semi-supervised anomaly detection has been adopted in several frameworks [166]. It leverages available annotated
normal footprints (having labels) and pertaining to the positive class to identify abnormalities from the negative class.
This is the case of deep autoencoder (DAE) architecture when it is only applied to learn normal consumption patterns
(with no anomalies). Accordingly, using enough training consumption observations from the normal category, the
autoencoder could generate low reconstruction errors for normal observations over abnormal patterns [167].
In [168], a semi-supervised support vector machine (semi-SVM) based anomaly detection solution is proposed, where a
small number of annotated power consumption patterns are required to train the learning model. This system can also
generate alarms if suspicious consumption patterns are detected, which are different to usual energy consumption habits
of the end-users. While in [169], DAE and ensemble k-nearest neighbor graphs (KNNG) are combined to develop a
semi-supervised anomaly detection system, in which only normal events with their labels are used to train the learning
model.

2.1.6 Other techniques (O)

In addition to what has been presented in the aforementioned subsections, there are other types of anomaly detection
techniques that are built on completely different strategies, including visualization and compressive sensing.

O1. Visualization: offers effective tools to comprehend consumption behavior of end-users through mapping consump-
tion footprints with visual spaces. In this line, visual experts make use of perceptual skills for helping end-users perceive
and decipher their consumption patterns within data. Moreover, visualization of load usage footprints could efficiently
aid in detecting anomalous consumption behaviors, faulty appliances and suspicious consumption fingerprints that may
be due to energy theft attacks. Accordingly, this enables end-users and energy managers to fix related issues and reduce
wasted energy.
For example, in [170], the authors propose an anomaly detection framework based on providing various time series
visualization schemes, which helps in analyzing and understanding energy consumption behavior. Moreover, it also
enables visualizing resulting anomaly scores to direct the end-user/analyst to important anomalous periods. In the same
way, an interactive visualization approach that helps in capturing power consumption anomalies is proposed in [171]. It
focuses on analyzing and visualizing spatio-temporal consumption footprints gleaned using various streaming data
sources. This method has been developed with respect to two prerequisites of real-world anomaly detection systems,
which are the online monitoring and interactivity. Moreover, an interactive dashboard is designed in [172] using an
early warning application, which can automatically analyze energy consumption footprints and provide end-users with
timely abnormal consumption visualizations based on data recorded from smart meters and sensors. While in [173], a
graphical visualization tool for supporting the detection and diagnosis of power consumption abnormalities using a
rule-based approach is proposed.

O2. Compressive sensing: represents a signal processing strategy for effectively analyzing and reconstructing time-
series data using their sparsity. It has been widely used in different research fields, such as facial recognition, holography
and monitoring of bio-signals. In addition, compressive sensing puts all the appropriate qualities to detect anomalies in
energy consumption [174]. For instance, in [175], the authors prove the relevance of applying compressive sensing in
sparse anomaly detection, it relies on the fact that the number of anomalous patterns is generally smaller than the total

12
 A PREPRINT - F EBRUARY 15, 2021

number of events. In the same direction, in [176], separable compression sensing is combined with PCA to identify
anomalous power data. In [177], anomalous events in smart grid are detected using a sparse approximation paradigm.

2.2 Anomaly detection level

The anomaly detection level of power consumption data plays a major role in developing effective solutions because it
describes either the level of resolution in which power anomalies have been detected and treated. Correspondingly,
tailored recommendations could be generated to resolve the associated issues and promote energy efficient behavior.

L1. Aggregated level: it refers to detecting anomalous power consumption using data of the main supply in a specific
building, i.e. without any information about individual consumption of the different appliances connected to the
electrical network. Although this kind of anomaly detection has been used in various works, it has the main drawback
of not being able to provide the end-user with information about which appliance is responsible for a specific anomaly.

L2. Appliance level: it stands for the case where anomaly detection is performed using appliance power consumption
data gathered using individual sub-meters. This kind of anomaly detection is widely adopted because it supports a
fine-grained tracking of abnormalities occurring during the operation of each electrical device [23].

L3. Spatio-temporal level: much attention has been devoted recently to the collection of continuous spatio-temporal
power consumption patterns from different devices and sources. This affords new opportunities to timely understand
consumption fingerprints in their spatio-temporal context [178, 179]. Overall, detecting anomalous consumption
behaviors using conventional data collection methods present considerable challenges since the boundary between
normal and anomalous observations is not obvious. Therefore, a straightforward solution to those challenges is to
interpret consumption abnormalities in their multifaceted and spatio-temporal context. Specifically, detecting abnormal
consumption related to specific hours in the day, or what are the severe days presenting anomalous consumption and
how to identify them in the timestamps (weekdays, weekends, holidays, etc.) will be valuable to provide end-users with
a personalized feedback to reduce wasted energy [180, 181].

2.3 Applications

The applications of anomaly detection of energy consumption in buildings are no longer limited to energy efficiency,
but they are finding themselves in various novel application contexts. Explicitly, they could be used for detecting (i)
abnormal consumption behaviors, (ii) faulty appliances, (iii) occupancy information, (iv) non-technical losses, and (v)
at-home elderly monitoring. In addition, the same anomaly detection system, within a building can be used for multiple
applications without the need for installing other systems (e.g. to detect occupancy or non-technical losses). Therefore,
this could effectively reduce the hardware implementation costs and decrease the complexity of installed systems.

A1. Detection of abnormal behavior of end-users: it is the main application for which anomaly detection has been
proposed since the final objective is to reduce wasted energy and promote sustainable and energy efficiency behaviors
[19, 150]. In this context, detecting anomalous consumption behavior of end-users allows a better and accurate
assessment of power usage, which can be translated into providing them with useful and personalized recommendations
[182, 183].

A2. Detection of faulty appliance: using various kinds of appliances at indoor environments has made people’s
lives more convenient. However, these electrical appliances could be faulty in different ways or could suffer from
inefficiencies, and hence leading to several issues, such as the events resulting in a massive energy waste and triggering
electrical fires [184, 185]. To that end, detecting faulty appliances and providing the end-users with customized
recommendations to replace them is of significant importance in reducing the operation cost and boosting energy saving
in buildings [23, 149].

A3. Occupancy detection: detecting whether a building or one of its parts is occupied by the end-users is essential to
allow a set of building automation tasks. Although actual tools for detecting the indoor occupancy typically need to
install specialized sensors, including passive-infrared sensors (PIR), reed switches actuated by magnets, or cameras,
their installation is very costly and further labor charges could be added for maintenance [186, 187]. Therefore, a
solution to overcome the high-cost pitfall is to explore the aptitude of electrical sub-meters, which are installed in most
of the houses around the globe to detect occupancy patterns [188, 189]. For example, the authors in [190] investigate
both appliance specific and aggregated load usage footprints to detect the occupancy of residents [191].

13
 A PREPRINT - F EBRUARY 15, 2021

A4. Non-technical loss detection: it mainly refers to (i) detecting unintentional sub-meters’ dysfunctions and electricity
theft attacks attempting to bypass sub-meters; (ii) braking and/or stopping sub-meters; (iii) identifying faulty sub-meters’
records; and (iv) capturing appliances having illegal connections [110, 192]. Non-technical loss in energy consumption
has negatively affected most of the economies over the globe [35]. For instance, more than 10% of produced energy
could be lost every year in Europe due to non-technical loss and billions of dollars are lost every year because of theft
energy attacks [149, 193]. To that end, detecting non-technical-loss and electricity theft have been introduced as an
information technology related challenge, which requires novel methods based on AI, data mining and forecasting
[106, 111]. Moreover, separating between behavioral consumption anomalies, fraud and unintentional consumption
deviations is reported as a current research trend to provide an accurate feedback to end-users and energy providers
[121, 151].

A5. At-home elderly monitoring: modern societies face significant issues with the monitoring of their elderly people
at home environments [194]. This problem could have considerable social and economic effects. However, one solution
to overcome it is via (i) monitoring appliance consumption of elderly people in real-time; (ii) identifying abnormal
consumption behaviors that could be occurring due to some critical situations (e.g. falls); and (iii) predicting faulty
operations of some appliances, which can result in dangerous situations (e.g. floods or gas leaks) [195, 196].

2.4 Computing platforms

As presented previously, most of the anomaly detection methods have been built upon the use of machine learning
techniques. However, although the use of these approaches has dived the development of anomaly detection technology,
it requires serious challenges of computing resources, data processing speed and scalability. In this regard, describing
and discussing available solutions used to implement anomaly detection systems is essential to understand the current
challenges.
• P1. Edge computing platforms: refer to distributed computational models that allow to drop the computing
resources and information storage capabilities close to the end-user application, where it can directly be used,
e.g. in energy consumption applications this can be done on the smart sensor platforms or smart plug devices,
as it is the case in (EM)3 [197]. Specifically, a smart plug is being developed to incorporate different sensors to
collect consumption and contextual data along with a micro-controller to pre-process data, segregate the main
consumption signal into device specific footprints, and detect abnormal behaviors. This helps in improving
output, accelerating data processing and saving bandwidth [198].
• P2. Fog computing platforms: stand for decentralized computational infrastructures, where data pre-
processing, computing, storage and analysis are conducted in the layer located between the data collection
devices and the cloud [199]. In this line, the computational ability of the anomaly detection solution is carried
out close to both the data recording devices and the cloud, in which data are produced and handled [200].
• P3. Cloud computing platforms: concern the cases when the computing and storage resources are ensured
using distant servers, in which the end-users deploying the anomaly detection solutions are required to connect
them through an internet link to be able to execute the anomaly detection algorithms [105]. Put differently, the
platforms used to implement these algorithms become as the access points for running the anomaly detection
applications and visualize the data held by the servers. The cloud architectures are described by their flexibility,
which allows the providers to constantly adjust the storage capability and computing power to the end-users’
requirements [201].
• P4. Hybrid computing platforms: refer to the cases where the computing power is guaranteed by various
layers, including the cloud, fog and edge as explained in [202]. In this context, based on the computing
requirement of the anomaly detection solution and the existing computing resources, the algorithms could
be executed either on the edge and/or fog when they need a low computation cost, otherwise they could be
implemented in the cloud when high computing cost is required [203, 204].
Table 1, presents a comparison of several aforementioned anomaly detection frameworks in building energy consumption.
They are compared with reference to various parameters, such as the (i) application scenario, (ii) category, (iii)
implemented technique, (iv) learning process, (v) computing platform used (or required) to implement the anomaly
detection algorithm, (vi) privacy preservation, and (vii) sampling rate. This helps in easily understanding the properties
of each framework and difference between existing solutions.

2.5 Example of anomaly detection using AI

In order to explain how anomalies of energy consumption have been considered in the literature and how AI could
be used to detect abnormal usage, we present in this section, three different scenarios for anomaly detection using (i)

14
 A PREPRINT - F EBRUARY 15, 2021

AI-based prediction, (ii) AI classification of energy micro-moments and occupancy data and (iii) one-class classification
of energy data. It is worth noting that with the use of AI, it becomes possible to detect more advanced kinds of anomalies
using other types of data, such as occupancy patterns and ambient conditions.

Scenario I. Anomaly detection using AI-based prediction:

In [205], a power consumption dataset is provided to validate anomaly detection algorithms. AI tools for predicting
future energy consumption are combined with a rule-based algorithm for detecting anomalous energy consumption.
This method relies on using an RNN model to predict energy consumption for the next timestamp and then the difference
between real and predicted consumption is calculated to measure the level of “surprise”. In this regard, if a significant
gap is detected, either (i) an anomalous energy usage behavior has occurred, or (ii) the model has made a mistake.
Moving forward, the investigation is then continued using a rule-based algorithm and by filtering the abnormalities
identified by the predictive system with reference to a set of statistical criteria. The rule-based algorithm helps in
detecting percentile of power consumption for each timestamp with regard to the (i) hour, (ii) outside temperature,
and (iii) type of day (working day vs. holiday). Fig. 4 illustrates an example of anomalies detected when analyzing
time-series energy consumption.

Scenario II. Anomaly detection using AI and micro-moment analysis:

In the (EM)3 framework [54], anomalous energy consumption is detected by analyzing energy consumption footprints
and occupancy patterns using micro-moments analysis. Following, a DNN model is deployed to automatically classify
each consumption observation as normal or abnormal. Specifically, energy consumption samples are clustered into five
classes; three of them are named “class 0: good usage”, “class 1: turn on a device” and “class 2: turn off a device”,
which represent normal usage; and the other two classes are called “class 3: excessive consumption” and “class 4:
consumption while outside”, which refer to abnormal usage. Fig. 5 illustrates an example of time-series energy traces
collected in the DRED dataset for the case of a television [206], and the corresponding normal and abnormal energy
patterns identified using a DNN model and micro-moment analysis. Because of occupancy data consideration in the
anomaly detection stage, it was possible to detect a new consumption anomaly that corresponds to the absence of the
end-user when the television is on (this abnormality could also be considered for other specific appliances, such as the
air conditioner, heater, fan, etc.), this was not possible using conventional anomaly detection techniques that are based
only on analyzing energy fingerprints.

15
 Table 1: Summary of research frameworks conducted in energy consumption anomaly detection.
Reference (year) Application Category Implemented technique Learning Computing Privacy Sampling
process platform preservation rate
[35] (2017) A1 U1 MNN and k-means clustering Unsupervised - - 1 hour
[45] (2020) A1 U2 OCRF Unsupervised P1 - 1 hour,
[54] (2020) A1 U2,U3 OCSVM, DBSCAN, LOF, LDA, IKNN Supervised P1,P3,P4 No 1 sec, 3 sec
[60] (2019) A1 S1 Autoencoder and RNN Unsupervised P3 - 1 min
[62] (2018) A1 S1 Variational recurrent autoencoder Supervised P3 - 15 min
[66] (2019) A4 S1 CNN and random forest Supervised P3 No 1 hour
[67] (2018) A4 S1 CNN Supervised P3 - 1 hour, 1 day
[75] (2020) A1 S1 Recurrent GAN Supervised P3 No 1 hour
[77] (2019) A1 S1 RNN and negative selection Supervised P2 No 30 min
[80] (2019) A1 S1 RNN and and K-means Unsupervised P3 - 1 hour
[81] (2020) A1 S1 RNN and quantile regression Unsupervised P4 - 1 hour
[85] (2020) A1 S1 ANNs and ARIMA Supervised P1 No 1 hour
[86] (2020) A2 S1 MLP Supervised P1,P2 - 1 hour
[22] (2019) A1 S2 Linear regression + rule-based algorithm Supervised P1 No 1 hour
[97] (2017) A1 S2 autoregressive prediction Semi-supervised P1,P2 - 30 min

16
[106] (2020) A4 S3 Bayes algorithms Supervised P1 No 5 min
[103] (2020) A4 S3 Bayesian networks Supervised P1 No 15 min
[105] (2016) A4 S3 Gaussian distribution Supervised P1 - 1 hour
[173] (2018) A1 O2 Graphical visualizatiuon Unsupervised P3 - 30 min, 1 hour
[111] (2019) A4 S4 SVM Supervised P1,P2 No 1 hour
[19] (2020) A1 S4,S1 SVM, KNN, decision tree, EBT, DNN Supervised P1,P3,P4 No 1/6 sec, 1 sec
[119] (2018) A1 E1 GBM Supervised P1 - 15 min
[121] (2019) A4 E1 GBM and grid search Supervised P1 - -
[125] (2017) A1 E2 Bootstrap aggregation Supervised P1,P2 - 5 min
[126] (2019) A1 E2 SCiForest Supervised P1 No 30 min
[133] (2016) A1 F1 Distance-based approach Unsupervised P1,P2 - -
[142] (2019) A1 F2 Time-series analysis Supervised P1,P2 No 1 min
[167] (2018) A1 H DAE Semi-supervised P3,P4 - 30 min
[168] (2019) A1 H Semi-SVM Semi-supervised P1 - 1 hour
[195] (2017) A5 F1 Rule-based algorithm Unsupervised P1 No 30 sec
[191] (2019) A4 F2, S1 Time-frequency features + OCRF Supervised P1,P2 - 10 min, 1 hour
[184] (2019) A2 S3 Rule based statistical model Unsupervised P1 No 10 min
[207] (2019) A4 S1 CNN Supervised P3 Yes 30 min
[208] (2019) A4 S1 CNN Supervised P3 Yes -
A PREPRINT - F EBRUARY 15, 2021
 A PREPRINT - F EBRUARY 15, 2021

Figure 4: Example of anomaly detection in time-series energy consumption using AI-based prediction applied on
anomaly detection dataset provided in [205].

Scenario III. Anomaly detection using one-class classification:

Another important anomaly detection solution is based on conventional one-class classification, which has been widely
utilized in other applications. Fig. 6 shows an example of an anomaly detection of energy consumption applied
on DRED dataset [206] using the one-class autoencoder. It has clearly been seen that this scheme divides energy
observation into two main classes based on the analysis of the energy consumption levels in a new representation space,
in which the power and time have been normalized [64].

3 Critical analysis and discussion

3.1 Discussion

Anomaly detection in building energy consumption is of paramount importance to developing powerful energy
management systems, identifying energy theft attacks, inefficiencies and negligence. However, in most cases it is
difficult to separate consumption abnormalities from the normal usage deviations occurring owing to seasonal changes
and variation of personal settings (e.g. holidays, family parties, unexpected changes of due new circumstances, etc.).
Moreover, one of the limitations of available anomaly detection methods is related to the fact that diverse unidentified
context data, including seasonal changes, could impact the power usage of end-users in a manner to be as abnormal
when existing time-series based anomaly detection techniques are used. In addition, a set of important findings can
summarized as follows:

17
 A PREPRINT - F EBRUARY 15, 2021

60

Power consum pt ion (wat t s)

50

40

30

20

10

07-05 00 07-05 03 07-05 06 07-05 09 07-05 12 07-05 15 07-05 18 07-05 21

Tim e & Dat e (m in)

Micro-moments
60 0
1
2
Power consumption (watts)

50 3
4
40

30

20

10

0
07-05 00 07-05 03 07-05 06 07-05 09 07-05 12 07-05 15 07-05 18 07-05 21 07-06 00
Time & Date (min)
Figure 5: Example of anomaly detection based on DNN and micro-moment analysis with reference to energy data and
occupancy patterns. These visualization plots are derived using DRED dataset (for the case of a television): time-series
energy consumption traces, and bottom) micro-moment detection scheme based on deep learning [54].

• AI-based solutions focus mainly on developing real-time or near real-time (e.g. at a hourly sampling rate or
lower) although they can also provide an insight analysis for time long periods (e.g. days, weeks, months and
years). This is due to the capability of AI to analyze big data, especially when high frequency sampling rates
are considered and also thanks to the IoT devices, smart-meters and smart sensors, which help tremendously
in collecting accurate data. On the other hand, this represents the main difference between actual AI-based
anomaly detection techniques and those used twenty or thirty years ago, where it was not possible to process
data in real-time or near real-time. In addition, almost all the reviewed frameworks have focused on the
analysis of power consumption data either on kWh or Wh. This depends on whether the anomaly detection
has been conducted at the aggregated-level (using kWh) or an appliance-level (using Wh).

• Most of existing approaches of anomaly detection in energy consumption attempt only to flag out power
samples that are remarkably higher or lower than usual consumption footprints, as it is the case in other
applications, such as bank card fraud detection, network intrusion detection and electrocardiogram anomaly
detection. Unfortunately, this is not the correct case to detect anomalous power consumption because the

18
 A PREPRINT - F EBRUARY 15, 2021

N ormal usage
Abnormal usage

0.8

N ormalized power consumption

0.6

0.4

0.2

0.0

0.0 0.2 0.4 0.6 0.8 1.0

N ormalized time

Figure 6: Example of anomaly detection in time-series energy consumption using autoencoders applied on DRED
dataset [206].

definition of anomaly in energy consumption can be quite different, other kinds of anomalies are available and
their detection requires other information sources, e.g. occupancy patterns, and appliance operation data.
• By using AI, it becomes possible to develop real-time or near real-time energy consumption anomaly detection
systems, which could identify timely anomalous usage and alert the end-users by sending warnings and
notifications. Accordingly, recommender systems could then be deployed to help the end-users with a better
decision-making to reduce their wasted energy through providing them with personalized and contextual
recommendations. For instance, the EM3 project 2 combines anomaly detection and a recommender system to
help end-users in reducing their wasted energy using both real-time or near real-time strategies.
• According to recent works [23, 20], using aggregated-level consumption data is not the best way to detect
anomalies of energy consumption because they are general and can not provide precise information on the
causes of each anomaly. Therefore, using appliance-level data generated either by sub-meters or using NILM
systems is more appropriate since this helps in detecting the anomalies of each appliance [22, 209].
• In some cases, the entirety of a given power consumption behavior could be considered as abnormal and not
only some specific observations, which make it difficult to detect the exact anomalous parts. Therefore, this
requires comparing current consumption footprints with the past and ideal consumption cycles and not only
using outlier detection algorithms, which can detect the anomalies at the sample level.
• In terms of the effectiveness of existing methods, although unsupervised anomaly detection is easy to implement
since it does not require annotated datasets to learn the anomalies, it presents serious drawbacks because
it can only detect one kind of anomalies, which is related to excessive consumption. This is also the same
with ensemble methods and feature extraction-based techniques. In contrast, supervised methods are not very
popular as unsupervised ones as they require using labeled datasets to learn the abnormalities. However, using
methods pertaining to this category allows to detect other types of anomalies. This is because they could
be defined a priori by human experts using training data collected from different sources, e.g. consumption
footprints, occupancy patterns, indoor conditions and appliance operation parameters.
• In terms of the computing resources, most of the deep learning based anomaly detection frameworks require
high-performance computing capabilities to conduct the learning process. Therefore, most of them use cloud
computing to integrate and manage large datasets. While for conventional machine learning based anomaly
detection, edge and fog computing have been successfully used in various frameworks and applications.
• Privacy preservation: developing anomaly detection systems to promote energy saving in buildings is of
paramount importance at all levels of the society. This can be performed using local and temporal fine-
2
http://em3.qu.edu.qa/

19
 A PREPRINT - F EBRUARY 15, 2021

grained records of power consumption fingerprints, occupancy patterns and ambient conditions to identify
abnormalities and unnecessary power consumption [210]. Unfortunately, using this kind of fine-grained
records enables disclosing information on the presence of the end-users based on their energy usage footprints.
In this context, we have noticed that the privacy preservation has been ignored or not reported in most of the
anomaly detection frameworks, only very few have tried to touch on this issue [207, 208].

3.2 Relevance of AI-based anomaly detection techniques

The relevance and robustness of AI-based anomaly detection does not rely only on the accuracy of detecting anomalous
energy usage, but also on the type and number of the consumption abnormalities that could be detected. In this regard,
it was clear that most of the unsupervised anomaly detection techniques (i.e. clustering, one-class classification and
dimensionality reduction) could detect only one type of energy usage anomalies, which corresponds to excessive energy
consumption. This is because they are based on identifying rare consumption observations or outliers, which raise
suspicions by differing considerably from the majority of the consumption footprints. In addition, they only analyze
energy consumption data without considering other relevant factors that impact energy usage, such as occupancy,
ambient conditions and users’ preferences. On the other hand, supervised anomaly detection presents more advantages
since they can be utilized to detect different kinds of energy consumption abnormalities by considering the impact of
the presence/absence of end-users, ambient conditions, outdoor weather data and users’ preferences on energy usage
[22, 64]. This was possible through the use of rule-based algorithms to define abnormal consumption and annotate
multi-modal datasets. In this context, deep anomaly detection techniques that are based on adopting deep learning
models presents promising performance and in terms of the accuracy of detecting abnormal usage and also because of
their capability to process and analyze multi-modal data, as described in [19]. Table 2 presents a summary of relevant
AI-based anomaly detection techniques, including their strengths and weaknesses.

3.3 Challenges and limitations

There are several common and domain-specific challenges and limitations of anomaly detection systems in energy
consumption, which hinder developing efficient solutions, render their implementation costly and limit their widespread.
They can be outlined in the following points:

• Absence of annotated datasets: among the serious pitfalls to develop and validate abnormality detection
schemes is the absence of annotated datasets, which provide labels for both normal and abnormal consumption.
Most of the supervised algorithms are validated on a small quantity of data, which can not be considered as
comprehensive datasets and are not accessible for the energy research community. Specifically, repositories
that label the events of abnormal consumption and their types almost do not exist and its creation is difficult and
costly [22]. Therefore, creating various datasets for different kinds of buildings that reflect real consumption
behaviors will help effectively the energy research community in testing and improving the detection of
consumption abnormalities in different application scenarios [211].
• Imbalanced dataset: refers to the distribution of anomalies through data classes, i.e. anomalous data might
usually be the minority amongst the overall dataset. Indeed, the anomaly data are very rare in reality, forming
together with the major normal data an extreme unbalanced set. The class imbalanced characteristic of most of
the anomaly detection datasets results in a sub-optimality of the algorithms’ performance. Therefore, to deal
with this issue, some pre-processing techniques are required, among them (i) using resampling procedures
to oversample the minority classes or undersample the majority classes, and (ii) generating synthetic power
consumption data [19]. Moreover, in other topics, the anomaly classes are generally represented as minor
classes, but in energy consumption this is not always the case, especially if a high energy wasting behavior is
observed. In this regards, applying unsupervised anomaly detection methods is less efficient.
• Definition of anomalies: traditional definition of an anomaly signifies that an anomalous observation is an
outlier or deviant. However, this definition could not be enough to define anomalies in energy consumption
because other forms of abnormalities could exist, e.g. keeping an appliance on (i.e. air conditioner, fan,
television, etc.) while end-users are outside, keeping windows and doors open when an air conditioner/heating
system is switching on, which leads to a high power consumption, etc. Therefore, to efficiently detect
anomalies of energy consumption, it is required to analyze not only the power consumption data but also other
information sources, including the occupancy patterns, ambient conditions, outside weather footprints and
appliance operation parameters.
• Sparse labels: on the one hand, the labels denoting whether an instance is normal or anomalous is in many
applications time-consuming and prohibitively expensive to obtain. This is especially typical for time series
data, where the sampling frequency could reach 1000 Hz or the time could range over decades, generating

20
 A PREPRINT - F EBRUARY 15, 2021

Table 2: A summary of relevant AI-based anomaly detection techniques, including their strengths and weaknesses.
Ref. Implemented technique Advantages Drawbacks
[35] MNN and k-means No need for annotated data Low detection accuracy, detection of only
clustering excessive consumption
[45] OCRF No need for annotated data Low detection accuracy, detection of only
excessive consumption
[54] OCSVM, DBSCAN, LOF Detection of Two kinds of anomalies, Absence of annotated data.
LOF, LDA, IKNN high detection accuracy
[60] Autoencoder and RNN No need for annotated data Detection of only excessive consumption,
low detection performance
[62] Variational recurrent No need for annotated data Difficulty to assess the performance
autoencoder
[66] CNN and random forest High anomaly detection Analyze only energy consumption data,
performance high computational cost
[67] CNN Capture the abnormal electricity Lack of annotated data, detection of excessive
usage with a high accuracy consumption, high computational cost
[77] RNN and negative Predict excessive consumption Low detection performance, only excessive
selection anomalies consumption is detected
[81] RNN and quantile High detection performance, no need Detection of only excessive consumption,
regression for annotated data weak interpretability
[85] ANNs and ARIMA Anomaly detection and energy usage High training cost, detection of only
prediction, high detection accuracy excessive consumption
[22] LR + rule-based algo Low training cost Difficulty to annotate data, low detection accuracy
[97] Autoregressive prediction Anomaly detection and power usage Low prediction performance, detection of only
prediction excessive consumption
[105] Gaussian distribution Low training cost Low detection performance, lack of annotated
data, detection of excessive consumption
[19] SVM, KNN, DT, Detection of Two kinds of anomalies, Lack of annotated data
EBT, DNN high detection performance
[119] GBM Anomaly detection and power usage Detection of only suspicious consumption
prediction levels, weak interpretability
[121] GBM and grid search Low training cost Low detection performance, weak interpretability,
one type of anomalies is detected
[125] Bootstrap aggregation High detection performance Difficulty to set the optimal threshold, detection
of only suspicious consumption level
[133] Distance-based approach Low training cost Weak interpretability, low detection performance

[142] Time-series analysis Low training cost Low detection accuracy, detection of only
excessive consumption
[167] DAE Anomaly detection and power usage High computation cost, detection of only
prediction, high detection accuracy excessive consumption
[168] Semi-SVM Anomaly detection and power usage Weak interpretability, detection of only
prediction suspicious consumption levels
[184] Rule-based statistical Anomalous appliances detection, Low detection accuracy, detection of only
model Low training cost excessive consumption
[207] CNN Privacy-preservation, high detection Detection of only suspicious consumption
accuracy levels, weak interpretability

an enormous amount of data points. On the other hand, anomalous data is often not reproducible and fully
concluded in reality.
• Detecting appliance-level anomalies is still not receiving the necessary attention, although it is more important
than detecting aggregated-level anomalies. In effect, a failure in the electronics of an appliance could not
only increase energy consumption, but in some cases, other kinds of failures may cause new forms of faulty
appliances that could be fatal, e.g. a faulty device can cause an electrical short that sparks a fire.
• Concept drift: this phenomenon usually occurs in time series data, where the common independent and
identically distributed (i.i.d) assumption for machine learning models is often violated due to the varying

21
 A PREPRINT - F EBRUARY 15, 2021

latent conditions [212]. Since the observations and relations in power consumption data evolve over time, they
should be analyzed near real-time, otherwise the systems implemented to analyze such data rapidly become
obsolete over time [213, 214]. In machine learning and data mining, this phenomenon is referred to as concept
drift.
• Absence of platforms to reproduce empirical results: one of the main issues of the anomaly detection in energy
consumption is the absence of platforms for reproducing the results of existing solutions. This may hinder the
performance comparison between existing algorithms and make it difficult to understand the state-of-the-art.
• Most of the frameworks differentiate between normal or abnormal power observations in general through
separating them into two principal classes (normal and abnormal) without further details. However, in real-
world scenarios, there exist different kinds of anomalous consumption, e.g. anomalies due to excessive
consumption of an appliance are different from those due to keeping a door of the refrigerator open or those
due to the absence of the end-user, as it is demonstrated in [54]. In this line, without providing the end-user
with the nature of anomalies and their sources, it is very difficult to trigger a behavioral change and promote
energy saving.

3.4 Market drivers and barriers

The frameworks reviewed in this article show that the anomaly detection topic is a promising strategy for a large
number of services and applications in the energy field. On the other hand, it is worth noting that the building energy
monitoring market in general, comprises a multi-billion USD global opportunity. This market appears to be growing
at a robust rate, in which the anomaly detection takes a significant part [215]. The decision-making of energy saving
systems in buildings depends on data, however, with the wide use of sub-meters and smart sensors, the data produced is
very huge which can frequently provoke the loss or misunderstanding of relevant information [216]. Various active
energy companies and utilities actually involved in providing anomaly detection and energy monitoring solutions,
markedly illustrate the increased importance of this technology to promote energy efficiency. Table 3 summarizes a
set of commercial anomaly detection of energy and energy management solutions developed by different companies,
which are used for different kinds of buildings. Specifically, it provides a description of each solution, company name,
frequency of energy monitoring and anomaly detection (real-time or near real-time), country and targeted building
environments.
In spite of the availability of the aforementioned solutions, different issues still require answers before enabling a
widespread deployment of the anomaly detection technology in the energy industry. First and foremost, anomaly
detection solutions should demonstrate that they could provide the scalability, speed and privacy preservation needed
for the considered application scenarios. Research efforts on distributed consensus algorithms, which are crucial to
achieving these objectives, are still ongoing, however a solution that combines all desired characteristics cannot yet
be achieved without significant trade-offs [230]. Albeit anomaly detection systems could be installed using existing
electric infrastructures, another crucial issue of these systems is that they have actually high implementation costs.
Most of the solutions are built upon the latest machine learning methods, which require high-performance computing
resources, e.g. using cloud platforms. Therefore, this slows down the commercialization of these solutions. Moreover,
resistance to security attacks resulting from unintentionally inappropriate system development or theft attacks are not
seriously addressed in most of the energy consumption anomaly detection solutions.

4 Current trends and new perspectives

After reviewing anomaly detection frameworks, discussing their limitation and drawbacks, and describing important
findings, it is of utmost importance to describe the current trends of this niche and derive the new perspectives that
could be targeted. This aids the anomaly detection community in understanding the current challenges and future
opportunities to improve the anomaly detection technology of energy consumption in buildings. Fig. 7 summarizes the
current trends and new perspectives that are identified in this framework.

4.1 Current trends

Anomaly detection in energy consumption presents various challenges, which are mainly domain-specific. For instance,
there is not a unique definition of normal versus anomalous consumption and there is inexplicit frontiers that separate
normal and anomalous behaviors. Moreover, there is an absence of ground-truth data and unified metrics that could be
deployed to evaluate the performance of anomaly detection algorithms. In addition, other data sources could result
in triggering non conventional energy consumption anomalies, such as: presence/absence of end-users, opening of

22
 A PREPRINT - F EBRUARY 15, 2021

Table 3: A summary of existing AI-based energy monitoring and anomaly detection commercial solutions for buildings.
Product Manufacturer Description Country Implementation
environment
Enetics SPEED [182] Enetics Faulty appliances identification and abnormal USA Public and domestic
consumption detection buildings
InBetween [217] Ecoisme Connected to the to the local Wi-Fi network and provides Poland Households
consumption statistics via mobile app
Informetis [218] Informetis Near real-time energy monitoring and analysis using IoT Japan Households
and big data mining technology
Verv Energy [219] Verv Energy Real-time electricity consumption monitoring with iOS UK Households
and Android app
Neurio [220] Neurio Real-time anomaly detection and notification Canada Households
(appliances ≥ 400 W)
WiBeee HOME [221] WiBeee Real-time consumption visualization, anomaly detection Spain Households
using cloud and energy saving recommendation
Smart Impulse [222] Smart Impulse Building’s energy consumption identification by end-use France Public buildings
(lighting, IT, heating, etc.) and anomaly detection
Verdigris [223] Verdigris Energy consumption monitoring and real-time fault USA Industrial and commercial
detection buildings
Voltaware [224] Voltaware Real-time energy monitoring and anomaly detection using UK Commercial, industrial
load disaggregation and tailored recommendations and domestic buildings
HOMEpulse [225] HOMEpulse Real-time energy disaggregation and anomaly detection France Households
(1-10 sec sampling rate)
Hive Starter Pack [226] AlertMe Electricity monitoring, appliances control using a mobile app UK Households
DiG Energy [227] Intelen Near real-time consumption monitoring, anomaly detection USA Commercial and domestic
End-users education about energy efficient practices buildings
Hark [228] Harksys Real-time anomaly detection and cost saving through UK Residential and public
buildings
EnerTalk [229] ENCORED Energy disaggregation, prediction and abnormal Korea Commercial and domestic
usage detection buildings

Figure 7: List of current trends and new perspective of anomaly detection in energy consumption.

23
 A PREPRINT - F EBRUARY 15, 2021

windows/doors when some specific appliances are on. To that end, this section discusses a set of current trends that
should be considered to enhance the anomaly detection technology for energy saving applications.

4.1.1 Considering other data sources

In traditional anomaly detection schemes deployed for energy consumption, the anomalies are generally detected using
only power consumption data gleaned from the main circuit or from individual devices, without paying any attention
to other factors that can affect the consumption. However, in order to conduct an accurate anomaly detection, all the
data that impact power consumption should be gleaned and stored along with energy consumption patterns. Following,
anomaly detection algorithms should be built with reference to all these data, which can be summarized as follows:

D1. Appliance parameters: each appliance has specific parameter settings that are responsible for its well functioning,
such as the minimum standby consumption, maximum standby consumption and maximum operation time. These
parameters are important to define normal and abnormal consumption of appliances and further to detect whether an
appliance is working perfectly or is faulty.

D2. Occupancy patterns: the presence or absence of end-users could highly affect energy usage and results in some
anomalous consumption behaviors that are not directly linked to excessive consumption of appliances. For example,
turning on an air conditioner, television, fan or desktop when end-users are absent should be considered as an abnormal
consumption behavior. To that end, recording occupancy data allows detecting unconventional anomalous consumption
behaviors.

D3. Ambient conditions: energy consumption could be extremely impacted by indoor conditions, such as the
temperature, humidity and luminosity since the operation of some appliances depends mainly on these factors (e.g. air
conditioners, heating systems, fans, light lamps, etc.). Therefore, collecting this kind of data aids in capturing abnormal
energy consumption.

4.1.2 Non-intrusive anomaly detection

Starting from the advantage of NILM as a good alternative to sub-metering for collecting itemized billing, its use
for detecting appliance-specific anomalies is very appreciated. Specifically, using NILM will remove the need to
install individual sub-meters for each appliance and hence helps in significantly reducing the cost of anomaly detection
solutions [18, 231]. The use of NILM to detect abnormal consumption results in the development of a new kind of
non-intrusive anomaly detection systems [232]. In [20, 233], the authors have attempted to investigate if device-specific
consumption fingerprints detected using NILM could be utilized directly to identify anomalous consumption behaviors
and to what extent this could impact the accuracy of the identification. Accordingly, even though the performance of
NILM to identify abnormal consumption is not yet as accurate as using sub-metering feedback, its performance could
be further improved to allow a robust identification of faulty behavior. Moving forward, more effort should be put in
this direction to develop non-intrusive anomaly detection of sufficient fidelity without the need to install additional
sub-meters [23, 234].

4.1.3 Collection of annotated datasets

As mentioned previously, the absence of annotated datasets impedes the development of power anomaly detection
solutions. To that end, greater effort should be put to collect and annotate power consumption datasets at different
building environments (households, workplaces, public buildings, and industrial buildings), and further to share them
publicly. This can help researchers to speed up the process of testing and validating their algorithms. In this context, the
authors in [19] launch two new datasets for anomaly detection. The former, called Qatar university dataset (QUD) is
collected in an energy lab and offers the consumption of four appliance categories along with the occupancy patterns for
a period of three months. While the latter, named power consumption simulated dataset (PCSiD), produces consumption
fingerprints of six devices and occupancy data for a period of two years. Both datasets provide power consumption
footprints with their associated labels, where the overall data is split into five consumption classes. Three of them
represent normal consumption classes, they are called “good consumption”, “turn on device” and “turn off device”,
while the two remaining classes refer to anomalous consumption groups, which are defined as “excessive consumption”
and “consumption while outside”. Fig. 8 resumes the assumption and labeling process of micro-moment classes, which
is applied in QUD and PCSiD 3 .

3
Both QUD and PCSiD datasets could be accessed via http://em3.qu.edu.qa/

24
 A PREPRINT - F EBRUARY 15, 2021

Figure 8: Micro-moments assumption and labeling used in [19] to cluster normal and abnormal energy consumption
data using a rule-based algorithm.

4.1.4 Unified metrics to measure the performance

In addition to what has been been presented and based on analyzing the state-of-the-art, it is worth mentioning that
there is no unified metrics and schemes to evaluate the performance of the anomaly detection algorithms. By contrast, a
fair comparison between different anomaly detection approaches should be conducted using an ensemble of standard
metrics, and should be performed under the same conditions, e.g. using the same dataset including appliance fingerprints
collected at the same sampling rate [22].

4.2 New perspectives

Recently, governments, end-users, utility companies and energy providers pay a significant interest to the anomaly
detection technology as a sustainable solution that could help in achieving the energy efficiency targets. In this section,
we provide a general overview of new perspectives in anomaly detection in energy consumption.

4.2.1 Explainable deep anomaly detection

Deep learning based based anomaly detection solutions receive an increasing attention in current frameworks. However,
despite their good performance, the black-box property of deep learning models represents a disadvantage in practical
implementation [235]. Particularly, in energy consumption anomaly detection schemes, explanations of abnormalities
detected using deep learning are critical. To that end, developing deep learning based abnormality detection techniques
providing explanations why a power consumption observation/event is abnormal supports end-users/experts in focusing
their investigations on the very crucial abnormalities and can boost their trust in the adopted solutions [236, 237].
For instance, one important orientation could be through developing a novel generation of explainable deep one-class
learning models to effectively detect different kinds of energy consumption anomalies [238]. Specifically, this category
of models helps in (i) learning a mapping to concentrate normal consumption observations in a feature space, (ii)
pushing abnormal patterns to be mapped away, and (iii) providing appropriate explanations for the anomalies detected,
or more exactly, a human-readable prescription presenting helpful information on the causes that have led to the anomaly.
Moreover, this enables generating tailored recommendations endorsing end-users to reducing their wasted energy and
energy providers to detecting non-technical losses through the use of explainable recommender systems (RS) [239].

4.2.2 Deep anomaly detection on the edge

Deep learning is one of the promising solutions to implement powerful anomaly detection solutions, however, a couple
of years ago, it had been pretended that deep learning could just be implemented on high-end computing platforms,
while the training/inference is conducted at the edge and carried out by edge servers, gateways or data centers. It had
been a legitimate presumption at that period since the tendency was through the distribution of computing resources

25
 A PREPRINT - F EBRUARY 15, 2021

Figure 9: Example of an edge-based anomaly detection solution used to develop a novel smart plug in the (EM)3
framework.

among the clouds and the edge serves. However, this situation is changed completely currently owing to recent R&D
achievements performed by academic and industrial partners [240]. Accordingly, the alternative considers the use
of novel microcontrollers that include integrated machine learning accelerators. This could bring machine learning
and specifically deep learning to the edge devices. The latter could not just execute machine learning algorithms, but
they do that while consuming very low power and they need to connect to cloud just if required. Overall, this kind of
microcontroller with embedded machine learning accelerators provides promising opportunities to offering computation
capability for energy sub-meters and sensors collecting ambient conditions (i.e. temperature, humidity and luminosity),
which gather data to enable various IoT applications [241].
On the other side, the edge is widely regarded as the furthest point in any IoT network that could be an advanced
gateway (or edge server). Furthermore, it terminates at the sub-meters/sensors near the end-user. Thus, placing more
analytical power near the end-user has become rational, where microcontrollers could be very convenient. Explicitly,
this allows the inference and eventually the training, to be performed on tiny and resource-constrained low-power
devices, instead of the large computing platforms (e.g. desktops, workstations, etc.) or the cloud. It is worth noting
that to implement deep learning models, their size needs to be reduced in order to adapt the moderate computing,
storage, and bandwidth resources of such devices, while maintaining the essential functionality and accuracy. Fig. 9
illustrates an example of the anomaly detection solution embedded on a microcontroller based smart plug, which is
under development in the (EM)3 project [242].

4.2.3 Deep reinforcement learning

Reinforcement learning is a promising topic of AI that has received a significant attention recently. Its concept is related
to comprehending the human decision-making procedure before developing algorithms for enabling agents to determine
the proper anomaly behavior using trial-and-error in parallel with the reception of feedback form of reward power
consumption signals [243]. In this regard, deep reinforcement learning (DRL) is then proposed as a merge of deep
learning and reinforcement learning to detect more complex consumption anomalies. Detecting such abnormalities
involves handling high-dimensional consumption patterns and environmental conditions, uncertainties of the agent’s
observations and sparse reward power consumption signatures. DRL techniques have been proposed lately to resolve a

26
 A PREPRINT - F EBRUARY 15, 2021

Figure 10: A novel architecture of visualization based anomaly detection in energy consumption to (i) improve the
accuracy of detected anomalies, and (ii) help end-users in comprehending their energy consumption footprints.

broad variety of issues, including detecting abnormalities video surveillance, traffic management and anomaly detection
[244, 245], communication and networking [246] and energy consumption prediction [247].
Overall, DRL shows promising opportunities to resolve effectively the problem of energy consumption anomaly
detection since the latter is considered as a decision-making task. Following, an agent is designed to learn from the
consumption and environmental data via a continuous interaction with them and reception of rewards for detected
anomalies, i.e. the process is similar to the natural human learning via their experiences.

4.2.4 Multimodal anomaly visualization

As explained previously, the capability to interpreting anomalous and normal power consumption behaviors is of
utmost importance since the essential intrinsic challenges in the abnormality detection issue are mainly related to (i) the
absence of obvious boundaries between anomalous and normal consumption observations and (ii) the complexity to
obtain annotated power consumption datasets to train and verify developed solutions. To that end, the knowledge and
experience of human experts are highly esteemed to judge the consumption scenarios. A subjective, comprehensive and
interactive visualization of power consumption patterns and resulted analytic is hence greatly helpful to support the
interpretation and facilitate an optimal decision-making. In this context, great attention has been devoted recently to
using innovative visualization tools and visual analysis methods to detect anomalous data in other research fields, such
as the spreading of rumors on social media [248] and user behavior [249, 250].
In this regard, using visualization and interactivity for detecting anomalous power consumption behaviors and supporting
end-users’ interpretability and interactivity represent a promising research direction, especially to understand sense-
making of anomalous consumption footprints and explain why an anomaly occurs. For instance, novel visualization
plots are designed in the (EM)3 framework to portray anomalous consumption patterns using a scatter plot, in which
two kind of anomalies, i.e. “excessive consumption” and “consumption without the presence of the end-user” along
with normal data are traced over the day time.
Furthermore, another notable visualization plot developed in (EM)3 , which could provide end-users with consumption
analytics and anomaly detection capabilities at an appliance-level is the stacked bar [251]. It enables to select devices
and stack various models of the same device altogether (e.g. televisions from distinct brands). Visualizing multi-level
power consumption could help end-users in effectively detecting anomalies and faulty devices, and hence could allow
them to perform better decision-making towards reducing wasted energy [252]. Fig. 10 portrays our perception of a
multimodal visualization based anomaly detection of energy consumption, in which visualization feedback (either at
the aggregated level or the appliance level) could be used to improve the accuracy of anomaly detection.

27
 A PREPRINT - F EBRUARY 15, 2021

4.2.5 Platforms for reproducible research

Despite the advance achieved in developing anomaly detection methods for energy consumption, principally, three
aspects affect reproducibility, and thus a fair and experimental comparison of anomaly detection algorithms: (i) it is
difficult to evaluate the generality anomaly detection techniques as most of the frameworks are generally assessed on a
unique dataset, (ii) there is an absence of frameworks comparing existing solutions under the same conditions, because
of the lack of available open-source anomaly detection datasets and (iii) distinct and standardized assessment criteria
are used in the state-of-the-art with regard to the considered scenario [253].
To overcome this issue, there is an urgent need to release an open source anomaly detection toolkit, which includes
challenging energy consumption datasets and existing anomaly detection algorithms. This will allow a fair and easy
comparison of anomaly detection algorithms in a reproducible manner. Furthermore, this will support the foundations
for future anomaly detection competitions [254].

4.2.6 Privacy-preserving machine learning

The wide use of machine learning methods for anomaly detection in energy consumption is actually limited by the lack
of open-access anomaly detection datasets to train and validate algorithms, due to strict legal and ethical requirements
to protect end-user privacy. Aiming at preserving end-user privacy while promoting scientific research while using
power consumption datasets, implementing novel approaches for federated, secure and privacy-preserving machine
learning is an urgent need. In this context, removing private information (anonymization) and replacing of vulnerable
inputs with artificially produced ones while permitting a reattribution based on a look up table (pseudonymization)
are among the solutions that could be targeted [255]. Furthermore, using federated machine learning, which helps in
training algorithms over various decentralized edge-devices/servers holding local power consumption patterns without
sharing them seems very promising for anomaly detection in energy consumption [256].

4.2.7 Explainable RS and the COVID-19 pandemic

Power consumption in buildings has been completely changed in the COVID-19 pandemic due to the constraints on
movement. This has widely triggered remote working and e-learning, and hence has shifted activities and energy
usage to domestic residents [257]. Therefore, the need for smart solutions to detect energy consumption anomalies
with reference to the actual situation and other changes that could be occurred at any time is a current challenge [258].
To that end, the use of RS for supporting human decision making has recently received increased interest [259, 260].
However, with the aim of increasing the end-user trust and improving the acceptance of the generated recommendations,
these systems should provide explanations [261].
In this context, developing mechanisms for explainable and persuasive energy consumption recommendations that
could be tailored based on the end-user preferences, habits and current circumstances will promptly reduce wasted
energy and promote energy saving [262]. Specifically, the explanations could justify the reasons for recommending
each energy efficiency act [263]. On the other hand, the persuasiveness of fact-based explanations could be improved
using persuasive and incentive aspects, such as emphasizing ecological impacts and economical saving benefits. Fig. 11
illustrates a general flowchart of an explainable energy RS proposed in the (EM)3 framework [4]. Moreover, it is worth
noting that explainable RS are much appropriate to unexpected energy consumption situations (e.g. the COVID-19
pandemic) since the recommendations could be generated in real-time in addition to providing the end-user with more
details (using contextual data) on each recommended action to increase its acceptance [264].

5 Conclusion
In this article, a systemic and technically-informed survey of anomaly detection methods in building energy consumption
has been presented. A taxonomy that classifies these approaches with reference to different aspects has been presented,
such as artificial intelligence models, application scenarios, detection level and computing platforms. To conclude,
anomaly detection strategies can evidently benefit energy saving systems, energy providers, end-users and governments
via reducing wasted consumption and energy costs. Specifically, they provide insight information on abnormal
consumption behavior, anomalous appliances, non-technical loss and electricity theft cyberattacks, but most significantly,
anomaly detection systems offer smart and powerful solutions for promoting energy saving. They also play a major role
in the energy monitoring market.
We have showed that the majority of anomaly detection solutions in energy consumption are still in their nascence.
To promote their widespread utilization and maturity, a set of challenges and limitations should be overcome, among
them the lack of annotated datasets, absence of the reproducibility platforms, and the lack of standard metrics to assess
the performance developed solutions. On the other hand, energy consumption is impacted by other factors such as,

28
 A PREPRINT - F EBRUARY 15, 2021

Figure 11: Example of the flowchart of an energy saving system based on the combination of anomaly detection and
RS, in which the output of the anomaly detection module serves as an input for the RS to help end-users in correcting
their energy consumption behaviors.

occupancy (presence/absence of the end-user), ambient conditions, outdoor temperatures and end-user’s preferences.
Therefore, it is of utmost importance to consider these data to develop powerful and reliable anomaly detection models,
which could detect more advance kinds of abnormal energy usage. All in all, a significant research effort should be
made in the near future to confront the aforementioned issues and improve the quality of anomaly detection systems.
In addition, further investigations are still ongoing in future directions, which could permit developing power anomaly
detection systems in terms of the scalability, decentralization, low power consumption, easy implementation and privacy
preservation. Finally, we believe that more research contributions, projects and collaborations with industrial partners
should be performed to help anomaly detection technology reach its entire potential, proving its commercial feasibility
and facilitating its mainstream adoption in residential buildings.

Acknowledgements

This paper was made possible by National Priorities Research Program (NPRP) grant No. 10-0130-170288 from the
Qatar National Research Fund (a member of Qatar Foundation). The statements made herein are solely the responsibility
of the authors.

References

[1] Y. Himeur, A. Alsalemi, F. Bensaali, A. Amira, C. Sardianos, I. Varlamis, G. Dimitrakopoulos, On the applicabil-
ity of 2d local binary patterns for identifying electrical appliances in non-intrusive load monitoring, in: K. Arai,
S. Kapoor, R. Bhatia (Eds.), Intelligent Systems and Applications, Springer International Publishing, Cham,
2021, pp. 188–205.

29
 A PREPRINT - F EBRUARY 15, 2021

[2] I. Varlamis, C. Sardianos, G. Dimitrakopoulos, A. Alsalemi, Y. Himeur, F. Bensaali, A. Amira, Reshaping

consumption habits by exploiting energy-related micro-moment recommendations: A case study, in: Communi-
cations in Computer and Information Science, Springer International Publishing, Cham, 2020, pp. 1–22.
[3] C. Sardianos, I. Varlamis, G. Dimitrakopoulos, D. Anagnostopoulos, A. Alsalemi, F. Bensaali, Y. Himeur,
A. Amira, Rehab-c: Recommendations for energy habits change, Future Generation Computer Systems 112
(2020) 394–407.
[4] C. Sardianos, I. Varlamis, C. Chronis, G. Dimitrakopoulos, A. Alsalemi, Y. Himeur, F. Bensaali, A. Amira, The
emergence of explainability of intelligent systems: Delivering explainable and personalized recommendations
for energy efficiency, International Journal of Intelligent Systems 36 (2) (2021) 656–680.
[5] H. Rau, P. Moran, R. Manton, J. Goggins, Changing energy cultures? household energy use before and after a
building energy efficiency retrofit, Sustainable Cities and Society 54 (2020) 101983.
[6] J. Ngarambe, G. Y. Yun, M. Santamouris, The use of artificial intelligence (ai) methods in the prediction of
thermal comfort in buildings: energy implications of ai-based thermal comfort controls, Energy and Buildings
211 (2020) 109807.
[7] Y. Himeur, A. Alsalemi, F. Bensaali, A. Amira, Efficient multi-descriptor fusion for non-intrusive appliance
recognition, in: 2020 IEEE International Symposium on Circuits and Systems (ISCAS), IEEE, 2020, pp. 1–5.
[8] A.-D. Pham, N.-T. Ngo, T. T. Ha Truong, N.-T. Huynh, N.-S. Truong, Predicting energy consumption in
multiple buildings using machine learning for improving energy efficiency and sustainability, Journal of Cleaner
Production 260 (2020) 121082.
[9] X. Luo, L. O. Oyedele, A. O. Ajayi, O. O. Akinade, Comparative study of machine learning-based multi-objective
prediction framework for multiple building energy loads, Sustainable Cities and Society 61 (2020) 102283.
[10] Y. Himeur, A. Alsalemi, A. Al-Kababji, F. Bensaali, A. Amira, Data fusion strategies for energy efficiency in
buildings: Overview, challenges and novel orientations, Information Fusion 64 (2020) 99 – 120.
[11] A. Alsalemi, C. Sardianos, F. Bensaali, I. Varlamis, A. Amira, G. Dimitrakopoulos, The role of micro-moments:
A survey of habitual behavior change and recommender systems for energy saving, IEEE Systems Journal 13 (3)
(2019) 3376–3387.
[12] C. fei Chen, G. Zarazua de Rubens, X. Xu, J. Li, Coronavirus comes home? energy use, home energy management,
and the social-psychological factors of covid-19, Energy Research & Social Science 68 (2020) 101688.
[13] C. Magazzino, M. Mele, N. Schneider, The relationship between air pollution and covid-19-related deaths: An
application to three french cities, Applied Energy 279 (2020) 115835.
[14] M. Brülisauer, L. Goette, Z. Jiang, J. Schmitz, R. Schubert, Appliance-specific feedback and social comparisons:
Evidence from a field experiment on energy conservation, Energy Policy 145 (2020) 111742.
[15] Y. Himeur, A. Alsalemi, F. Bensaali, A. Amira, Improving in-home appliance identification using fuzzy-neighbors-
preserving analysis based qr-decomposition, in: International Congress on Information and Communication
Technology, Springer, 2020, pp. 303–311.
[16] Y. Himeur, A. Alsalemi, F. Bensaali, A. Amira, Robust event-based non-intrusive appliance recognition using
multi-scale wavelet packet tree and ensemble bagging tree, Applied Energy 267 (2020) 114877.
[17] A. Elsalemi, Y. Himeur, F. Bensaali, A. Amira, Appliance-level monitoring with micro-moment smart plugs, in:
The Fifth International Conference on Smart City Applications (SCA), 2020, pp. 1–5.
[18] H. Rashid, P. Singh, Monitor: An abnormality detection approach in buildings energy consumption, in: 2018
IEEE 4th International Conference on Collaboration and Internet Computing (CIC), 2018, pp. 16–25.
[19] Y. Himeur, A. Alsalemi, F. Bensaali, A. Amira, A novel approach for detecting anomalous energy consumption
based on micro-moments and deep neural networks, Cognitive Computation 12 (6) (2020) 1381–1401.
[20] H. Rashid, P. Singh, V. Stankovic, L. Stankovic, Can non-intrusive load monitoring be used for identifying an
appliance’s anomalous behaviour?, Applied Energy 238 (2019) 796 – 805.
[21] A. Wang, J. C. Lam, S. Song, V. O. Li, P. Guo, Can smart energy information interventions help householders
save electricity? a svr machine learning approach, Environmental Science & Policy 112 (2020) 381 – 393.
[22] M. Gaur, S. Makonin, I. V. Bajić, A. Majumdar, Performance evaluation of techniques for identifying abnormal
energy consumption in buildings, IEEE Access 7 (2019) 62721–62733.
[23] H. Rashid, V. Stankovic, L. Stankovic, P. Singh, Evaluation of non-intrusive load monitoring algorithms for
appliance-level anomaly detection, in: ICASSP 2019 - 2019 IEEE International Conference on Acoustics, Speech
and Signal Processing (ICASSP), 2019, pp. 8325–8329.

30
 A PREPRINT - F EBRUARY 15, 2021

[24] K. S. A. Kumar, A. M. M. O. Chacko, Clustering algorithms for intrusion detection: A broad visualization,
ICTCS ’16, Association for Computing Machinery, New York, NY, USA, 2016.
[25] R. A. Ariyaluran Habeeb, F. Nasaruddin, A. Gani, M. A. Amanullah, I. Abaker Targio Hashem, E. Ahmed,
M. Imran, Clustering-based real-time anomaly detection—a breakthrough in big data technologies, Transactions
on Emerging Telecommunications Technologies (2019) 1–27.
[26] E. Vanem, A. Brandsæter, Unsupervised anomaly detection based on clustering methods and sensor data on a
marine diesel engine, Journal of Marine Engineering & Technology 0 (0) (2019) 1–18.
[27] K. Verma, B. Singh, A. Dixit, A review of supervised and unsupervised machine learning techniques for suspi-
cious behavior recognition in intelligent surveillance system, International Journal of Information Technology
(2019) 1–14.
[28] M. Ahmed, A. N. Mahmood, M. R. Islam, A survey of anomaly detection techniques in financial domain, Future
Generation Computer Systems 55 (2016) 278 – 288.
[29] B. Feng, Q. Li, X. Pan, J. Zhang, D. Guo, Groupfound: An effective approach to detect suspicious accounts in
online social networks, International Journal of Distributed Sensor Networks 13 (7) (2017) 1550147717722499.
[30] V. Pastore, T. Zimmerman, S. Biswas, S. Bianco, Annotation-free learning of plankton for classification and
anomaly detection, Scientific Reports 10 (2020) 1–15.
[31] P. Arjunan, H. D. Khadilkar, T. Ganu, Z. M. Charbiwala, A. Singh, P. Singh, Multi-user energy consumption
monitoring and anomaly detection with partial context information, in: Proceedings of the 2nd ACM International
Conference on Embedded Systems for Energy-Efficient Built Environments, BuildSys ’15, Association for
Computing Machinery, New York, NY, USA, 2015, p. 35–44.
[32] B. Rossi, S. Chren, B. Buhnova, T. Pitner, Anomaly detection in smart grid data: An experience report, in: 2016
IEEE International Conference on Systems, Man, and Cybernetics (SMC), 2016, pp. 002313–002318.
[33] J. Henriques, F. Caldeira, T. Cruz, P. Simões, Combining k-means and xgboost models for anomaly detection
using log datasets, Electronics 9 (7) (2020) 1–16.
[34] H. Izakian, W. Pedrycz, Anomaly detection in time series data using a fuzzy c-means clustering, in: 2013 Joint
IFSA World Congress and NAFIPS Annual Meeting (IFSA/NAFIPS), 2013, pp. 1513–1518.
[35] J. Yeckle, B. Tang, Detection of electricity theft in customer consumption using outlier detection algorithms, in:
2018 1st International Conference on Data Intelligence and Security (ICDIS), 2018, pp. 135–140.
[36] G. Nychis, V. Sekar, D. G. Andersen, H. Kim, H. Zhang, An empirical evaluation of entropy-based traffic
anomaly detection, IMC ’08, Association for Computing Machinery, New York, NY, USA, 2008, p. 151–156.
[37] P. Bereziński, B. Jasiul, M. Szpyrka, An entropy-based network anomaly detection method, Entropy 17 (4)
(2015) 2367–2408.
[38] Z. Shi, P. Li, Y. Sun, An outlier generation approach for one-class random forests: An example in one-
class classification of remote sensing imagery, in: 2016 IEEE International Geoscience and Remote Sensing
Symposium (IGARSS), 2016, pp. 5107–5110.
[39] L. Ruff, R. Vandermeulen, N. Goernitz, L. Deecke, S. A. Siddiqui, A. Binder, E. Müller, M. Kloft, Deep one-class
classification, Vol. 80 of Proceedings of Machine Learning Research, PMLR, Stockholmsmässan, Stockholm
Sweden, 2018, pp. 4393–4402.
[40] V. R. Jakkula, D. J. Cook, Detecting anomalous sensor events in smart home data for enhancing the living
experience., in: Artificial Intelligence and Smarter Living, Vol. WS-11-07 of AAAI Workshops, AAAI, 2011.
[41] R. Chalapathy, A. K. Menon, S. Chawla, Anomaly detection using one-class neural networks (2018). arXiv:
1802.06360.
[42] P. Oza, V. M. Patel, One-class convolutional neural network, IEEE Signal Processing Letters 26 (2) (2019)
277–281.
[43] M. Zhang, J. Wu, H. Lin, P. Yuan, Y. Song, The application of one-class classifier based on cnn in image defect
detection, Procedia Computer Science 114 (2017) 341 – 348, complex Adaptive Systems Conference with
Theme: Engineering Cyber Physical Systems, CAS October 30 – November 1, 2017, Chicago, Illinois, USA.
[44] C. Désir, S. Bernard, C. Petitjean, L. Heutte, One class random forests, Pattern Recognition 46 (12) (2013) 3490
– 3506.
[45] K. Ghori, M. Imran, A. Nawaz, R. Abbasi, A. Ullah, L. Szathmary, Performance analysis of machine learning
classifiers for non-technical loss detection, Journal of Ambient Intelligence and Humanized Computing (2020)
1–16doi:10.1007/s12652-019-01649-9.

31
 A PREPRINT - F EBRUARY 15, 2021

[46] F. T. Liu, K. M. Ting, Z.-H. Zhou, Isolation-based anomaly detection, ACM Transactions on Knowledge
Discovery from Data (TKDD) 6 (1) (2012) 1–39.
[47] T. Huang, H. Sethu, N. Kandasamy, A new approach to dimensionality reduction for anomaly detection in data
traffic, IEEE Transactions on Network and Service Management 13 (3) (2016) 651–665.
[48] M. Valko, B. Kveton, H. Valizadegan, G. F. Cooper, M. Hauskrecht, Conditional anomaly detection with soft
harmonic functions, in: 2011 IEEE 11th International Conference on Data Mining, 2011, pp. 735–743.
[49] P. Naveen, W. K. Ing, M. K. Danquah, A. S. Sidhu, A. Abu-Siada, Cloud computing for energy management
in smart grid - an application survey, IOP Conference Series: Materials Science and Engineering 121 (2016)
012010.
[50] C. E. Brown, Multiple Discriminant Analysis, Springer Berlin Heidelberg, Berlin, Heidelberg, 1998, pp. 115–128.
[51] J. Wu, X. Zhang, A pca classifier and its application in vehicle detection, in: IJCNN’01. International Joint
Conference on Neural Networks. Proceedings (Cat. No. 01CH37222), Vol. 1, IEEE, 2001, pp. 600–604.
[52] T. Kudo, T. Morita, T. Matsuda, T. Takine, Pca-based robust anomaly detection using periodic traffic behavior,
in: 2013 IEEE International Conference on Communications Workshops (ICC), 2013, pp. 1330–1334. doi:
10.1109/ICCW.2013.6649443.
[53] A. Sial, A. Singh, A. Mahanti, Detecting anomalous energy consumption using contextual analysis of smart
meter data, Wireless Networks (2019) 1–18.
[54] Y. Himeur, A. Elsalemi, F. Bensaali, A. Amira, Smart power consumption abnormality detection in buildings
using micro-moments and improved k-nearest neighbors, Intenational Journal of Intelligent Systems (2021)
1–25.
[55] K. Kamaraj, B. Dezfouli, Y. Liu, Edge mining on iot devices using anomaly detection, in: 2019 Asia-Pacific
Signal and Information Processing Association Annual Summit and Conference (APSIPA ASC), 2019, pp.
33–40.
[56] K. M. Alheeti, A. Gruebler, K. McDonald-Maier, Using discriminant analysis to detect intrusions in external
communication for self-driving vehicles, Digital Communications and Networks 3 (3) (2017) 180 – 187.
[57] M. Chijoriga, Application of multiple discriminant analysis (mda) as a credit scoring and risk assessment model,
International Journal of Emerging Markets 6 (2011) 132–147.
[58] B. R. Kiran, D. M. Thomas, R. Parakkal, An overview of deep learning based methods for unsupervised and
semi-supervised anomaly detection in videos, Journal of Imaging 4 (2) (2018) 1–25.
[59] R. Wang, K. Nie, T. Wang, Y. Yang, B. Long, Deep learning for anomaly detection, WSDM ’20, Association for
Computing Machinery, New York, NY, USA, 2020, p. 894–896.
[60] Y. Weng, N. Zhang, C. Xia, Multi-agent-based unsupervised detection of energy consumption anomalies on
smart campus, IEEE Access 7 (2019) 2169–2178.
[61] X. Wang, T. Zhao, H. Liu, R. He, Power consumption predicting and anomaly detection based on long short-term
memory neural network, in: 2019 IEEE 4th International Conference on Cloud Computing and Big Data Analysis
(ICCCBDA), 2019, pp. 487–491.
[62] J. Pereira, M. Silveira, Unsupervised anomaly detection in energy time series data using variational recurrent
autoencoders with attention, in: 2018 17th IEEE International Conference on Machine Learning and Applications
(ICMLA), 2018, pp. 1275–1282.
[63] Y. Yuan, K. Jia, A distributed anomaly detection method of operation energy consumption using smart meter
data, in: 2015 International Conference on Intelligent Information Hiding and Multimedia Signal Processing
(IIH-MSP), 2015, pp. 310–313.
[64] Y. Himeur, A. Elsalemi, F. Bensaali, A. Amira, Detection of appliance-level abnormal energy consumption
in buildings using autoencoders and micro-moments, in: The Fifth International Conference on Big Data and
Internet of Things (BDIoT), 2021, pp. 1–13.
[65] M. Munir, S. A. Siddiqui, M. A. Chattha, A. Dengel, S. Ahmed, Fusead: Unsupervised anomaly detection in
streaming sensors data by fusing statistical and deep learning models, Sensors 19 (11).
[66] S. Li, Y. Han, X. Yao, S. Yingchen, J. Wang, Q. Zhao, P.-F. Pai, Electricity theft detection in power grids with
deep learning and random forests, Journal of Electrical and Computer Engineering 2019 (2019) 1–12.
[67] Z. Zheng, Y. Yang, X. Niu, H. Dai, Y. Zhou, Wide and deep convolutional neural networks for electricity-theft
detection to secure smart grids, IEEE Transactions on Industrial Informatics 14 (4) (2018) 1606–1615.

32
 A PREPRINT - F EBRUARY 15, 2021

[68] Z. Tang, Z. Chen, Y. Bao, H. Li, Convolutional neural network-based data anomaly detection method using
multiple information for structural health monitoring, Structural Control and Health Monitoring 26 (1) (2019)
e2296, e2296 STC-18-0112.R1.
[69] C. Zhang, D. Song, Y. Chen, X. Feng, C. Lumezanu, W. Cheng, J. Ni, B. Zong, H. Chen, N. Chawla, A deep
neural network for unsupervised anomaly detection and diagnosis in multivariate time series data (11 2018).
[70] K. Alrawashdeh, C. Purdy, Toward an online anomaly intrusion detection system based on deep learning, in:
2016 15th IEEE International Conference on Machine Learning and Applications (ICMLA), 2016, pp. 195–200.
[71] Y. Choi, H. Lim, H. Choi, I. Kim, Gan-based anomaly detection and localization of multivariate time series data
for power plant, in: 2020 IEEE International Conference on Big Data and Smart Computing (BigComp), 2020,
pp. 71–74.
[72] Y. Sun, W. Yu, Y. Chen, A. Kadam, Time series anomaly detection based on gan, in: 2019 Sixth International
Conference on Social Networks Analysis, Management and Security (SNAMS), 2019, pp. 375–382.
[73] D. Li, D. Chen, B. Jin, L. Shi, J. Goh, S.-K. Ng, Mad-gan: Multivariate anomaly detection for time series data
with generative adversarial networks, in: I. V. Tetko, V. Kůrková, P. Karpov, F. Theis (Eds.), Artificial Neural
Networks and Machine Learning – ICANN 2019: Text and Time Series, Springer International Publishing, Cham,
2019, pp. 703–716.
[74] S. Huang, K. Lei, Igan-ids: An imbalanced generative adversarial network towards intrusion detection system in
ad-hoc networks, Ad Hoc Networks 105 (2020) 102177.
[75] M. N. Fekri, A. M. Ghosh, K. Grolinger, Generating energy data for machine learning with recurrent generative
adversarial networks, Energies 13 (1).
[76] L. Bontemps, V. L. Cao, J. McDermott, N.-A. Le-Khac, Collective anomaly detection based on long short-term
memory recurrent neural networks, in: T. K. Dang, R. Wagner, J. Küng, N. Thoai, M. Takizawa, E. Neuhold
(Eds.), Future Data and Security Engineering, Springer International Publishing, Cham, 2016, pp. 141–152.
[77] A. daSilva, I. S. Guarany, B. Arruda, E. C. Gurjão, R. S. Freire, A method for anomaly prediction in power
consumption using long short-term memory and negative selection, in: 2019 IEEE International Symposium on
Circuits and Systems (ISCAS), 2019, pp. 1–5.
[78] K. Hollingsworth, K. Rouse, J. Cho, A. Harris, M. Sartipi, S. Sozer, B. Enevoldson, Energy anomaly detection
with forecasting and deep learning, in: 2018 IEEE International Conference on Big Data (Big Data), 2018, pp.
4921–4925.
[79] G. Fenza, M. Gallo, V. Loia, Drift-aware methodology for anomaly detection in smart grid, IEEE Access 7
(2019) 9645–9657.
[80] C. Chahla, H. Snoussi, L. Merghem-Boulahia, M. Esseghir, A Novel Approach for Anomaly Detection in Power
Consumption Data, in: 8th International Conference on Pattern Recognition Applications and Methods, Prague,
Czech Republic, 2019.
[81] C. Xu, H. Chen, Abnormal energy consumption detection for gshp system based on ensemble deep learning and
statistical modeling method, International Journal of Refrigeration 114 (2020) 106–117.
[82] H. Chen, X. Fei, S. Wang, X. Lu, G. Jin, W. Li, X. Wu, Energy consumption data based machine anomaly
detection, in: 2014 Second International Conference on Advanced Cloud and Big Data, 2014, pp. 136–142.
[83] A. Santolamazza, V. Cesarotti, V. Introna, Anomaly detection in energy consumption for condition-based mainte-
nance of compressed air generation systems: an approach based on artificial neural networks, IFAC-PapersOnLine
51 (11) (2018) 1131 – 1136, 16th IFAC Symposium on Information Control Problems in Manufacturing INCOM
2018.
[84] M. Ghanbari, W. Kinsner, K. Ferens, Anomaly detection in a smart grid using wavelet transform, variance fractal
dimension and an artificial neural network, in: 2016 IEEE Electrical Power and Energy Conference (EPEC),
2016, pp. 1–6.
[85] X. Wang, S.-H. Ahn, Real-time prediction and anomaly detection of electrical load in a residential community,
Applied Energy 259 (2020) 114145.
[86] J. Mulongo, M. Atemkeng, T. Ansah-Narh, R. Rockefeller, G. M. Nguegnang, M. A. Garuti, Anomaly detection
in power generation plants using machine learning and neural networks, Applied Artificial Intelligence 34 (1)
(2020) 64–79.
[87] L. Van Efferen, A. M. T. Ali-Eldin, A multi-layer perceptron approach for flow-based anomaly detection, in:
2017 International Symposium on Networks, Computers and Communications (ISNCC), 2017, pp. 1–6.

33
 A PREPRINT - F EBRUARY 15, 2021

[88] K. Kammerer, B. Hoppenstedt, R. Pryss, S. Stökler, J. Allgaier, M. Reichert, Anomaly detections for manufactur-
ing systems based on sensor data—insights into two challenging real-world production settings, Sensors 19 (24)
(2019) 1–18.
[89] Y. Zeng, J. Zhuang, Construction cosine radial basic function neural networks based on artificial immune
networks, in: L. Cao, J. Zhong, Y. Feng (Eds.), Advanced Data Mining and Applications, Springer Berlin
Heidelberg, Berlin, Heidelberg, 2010, pp. 134–141.
[90] V. M. Janakiraman, D. Nielsen, Anomaly detection in aviation data using extreme learning machines, in: 2016
International Joint Conference on Neural Networks (IJCNN), 2016, pp. 1993–2000.
[91] S. K. Bose, B. Kar, M. Roy, P. K. Gopalakrishnan, A. Basu, Adepos: Anomaly detection based power saving for
predictive maintenance using edge computing (2018). arXiv:1811.00873.
[92] Y. Imamverdiyev, L. Sukhostat, Anomaly detection in network traffic using extreme learning machine, in: 2016
IEEE 10th International Conference on Application of Information and Communication Technologies (AICT),
2016, pp. 1–4.
[93] R. Kromanis, P. Kripakaran, Support vector regression for anomaly detection from measurement histories,
Advanced Engineering Informatics 27 (4) (2013) 486 – 495.
[94] Y. Zhang, W. Chen, J. Black, Anomaly detection in premise energy consumption data, in: 2011 IEEE Power and
Energy Society General Meeting, 2011, pp. 1–8.
[95] M. Fahim, A. Sillitti, An anomaly detection model for enhancing energy management in smart buildings, in:
2018 IEEE International Conference on Communications, Control, and Computing Technologies for Smart Grids
(SmartGridComm), 2018, pp. 1–6.
[96] X. Liu, P. S. Nielsen, Scalable prediction-based online anomaly detection for smart meter data, Information
Systems 77 (2018) 34 – 47.
[97] W. Cui, H. Wang, A new anomaly detection system for school electricity consumption data, Information 8 (4).
[98] J.-S. Chou, A. S. Telaga, Real-time detection of anomalous power consumption, Renewable and Sustainable
Energy Reviews 33 (2014) 400 – 411.
[99] D. Bacciu, P. J. Lisboa, A. Sperduti, T. Villmann, Probabilistic Modeling in Machine Learning, Springer Berlin
Heidelberg, Berlin, Heidelberg, 2015, pp. 545–575.
[100] E. Roberts, B. A. Bassett, M. Lochner, Bayesian anomaly detection and classification for noisy data, in:
A. Abraham, P. Siarry, K. Ma, A. Kaklauskas (Eds.), Intelligent Systems Design and Applications, Springer
International Publishing, Cham, 2021, pp. 426–435.
[101] H. N. Akouemo, R. J. Povinelli, Probabilistic anomaly detection in natural gas time series data, International
Journal of Forecasting 32 (3) (2016) 948 – 956.
[102] L. Rashidi, S. Hashemi, A. Hamzeh, Anomaly detection in categorical datasets using bayesian networks, in:
Artificial Intelligence and Computational Intelligence, Vol. 7003, 2011, pp. 610–619.
[103] S. Saqaeeyan, H. Haj Seyyed Javadi, H. Amirkhani, Anomaly detection in smart homes using bayesian networks,
KSII Transactions on Internet and Information Systems 14 (2020) 1796–1816.
[104] V. Jakkula, D. Cook, Outlier detection in smart environment structured power datasets, in: 2010 Sixth Interna-
tional Conference on Intelligent Environments, 2010, pp. 29–33.
[105] X. Liu, N. Iftikhar, P. S. Nielsen, A. Heller, Online anomaly energy consumption detection using lambda
architecture, in: S. Madria, T. Hara (Eds.), Big Data Analytics and Knowledge Discovery, Springer International
Publishing, Cham, 2016, pp. 193–209.
[106] D. Hock, M. Kappes, B. Ghita, Using multiple data sources to detect manipulated electricity meter by an
entropy-inspired metric, Sustainable Energy, Grids and Networks 21 (2020) 100290.
[107] B. Coma-Puig, J. Carmona, R. Gavaldà, S. Alcoverro, V. Martin, Fraud detection in energy consumption: A
supervised approach, in: 2016 IEEE International Conference on Data Science and Advanced Analytics (DSAA),
2016, pp. 120–129.
[108] D. Janakiram, A. V. U. P. Kumar, A. M. Reddy V., Outlier detection in wireless sensor networks using bayesian
belief networks, in: 2006 1st International Conference on Communication Systems Software Middleware, 2006,
pp. 1–6.
[109] B. Chen, M. Sinn, J. Ploennigs, A. Schumann, Statistical anomaly detection in mean and variation of energy
consumption, in: Proceedings of the 2014 22nd International Conference on Pattern Recognition, ICPR ’14,
IEEE Computer Society, USA, 2014, p. 3570–3575.

34
 A PREPRINT - F EBRUARY 15, 2021

[110] S. S. S. R. Depuru, L. Wang, V. Devabhaktuni, Support vector machine based data classification for detection of
electricity theft, in: 2011 IEEE/PES Power Systems Conference and Exposition, 2011, pp. 1–8.
[111] A. Amara korba, N. El Islem karabadji, Smart grid energy fraud detection using svm, in: 2019 International
Conference on Networking and Advanced Systems (ICNAS), 2019, pp. 1–6.
[112] J. Nagi, K. S. Yap, S. K. Tiong, S. K. Ahmed, A. M. Mohammad, Detection of abnormalities and electricity theft
using genetic support vector machines, in: TENCON 2008 - 2008 IEEE Region 10 Conference, 2008, pp. 1–6.
[113] L. Zhang, L. Wan, Y. Xiao, S. Li, C. Zhu, Anomaly detection method of smart meters data based on gmm-lda
clustering feature learning and pso support vector machine, in: 2019 IEEE Sustainable Power and Energy
Conference (iSPEC), 2019, pp. 2407–2412.
[114] C. Cody, V. Ford, A. Siraj, Decision tree learning for fraud detection in consumer energy consumption, in: 2015
IEEE 14th International Conference on Machine Learning and Applications (ICMLA), 2015, pp. 1175–1179.
[115] M. Reif, M. Goldstein, A. Stahl, T. M. Breuel, Anomaly detection by combining decision trees and parametric
densities, in: 2008 19th International Conference on Pattern Recognition, 2008, pp. 1–4.
[116] X. Xu, H. Liu, M. Yao, Recent progress of anomaly detection, Complexity 2019 (2019) 1–11.
[117] E. Bahri, N. Harbi, H. N. Huu, Approach based ensemble methods for better and faster intrusion detection, in:
Á. Herrero, E. Corchado (Eds.), Computational Intelligence in Security for Information Systems, Springer Berlin
Heidelberg, Berlin, Heidelberg, 2011, pp. 17–24.
[118] J. Zhang, Z. Wang, J. Meng, Y. Tan, J. Yuan, Boosting positive and unlabeled learning for anomaly detection
with multi-features, IEEE Transactions on Multimedia 21 (5) (2019) 1332–1344.
[119] S. Touzani, J. Granderson, S. Fernandes, Gradient boosting machine for modeling the energy consumption of
commercial buildings, Energy and Buildings 158 (2018) 1533 – 1543.
[120] B. Adhi Tama, K. H. Rhee, An in-depth experimental study of anomaly detection using gradient boosted machine,
Neural Computing and Applications 31 (2019) 955–965.
[121] B. Albiero, R. Santos, E. Uyrá, R. Vilarino, J. Silva, T. Souza, R. Vicente, S. Yamouni, Employing gradient
boosting and anomaly detection for prediction of frauds in energy consumption, in: Anais do XVI Encontro
Nacional de Inteligência Artificial e Computacional, SBC, Porto Alegre, RS, Brasil, 2019, pp. 916–925.
[122] T. Kim, D. Lee, J. Choi, A. Spurlock, A. Sim, A. Todd, K. Wu, Extracting baseline electricity usage using gradient
tree boosting, in: 2015 IEEE International Conference on Smart City/SocialCom/SustainCom (SmartCity), 2015,
pp. 734–741.
[123] D. P. Gaikwad, R. C. Thool, Intrusion detection system using bagging ensemble method of machine learning, in:
2015 International Conference on Computing Communication Control and Automation, 2015, pp. 291–295.
[124] H. V. Nguyen, H. H. Ang, V. Gopalkrishnan, Mining outliers with ensemble of heterogeneous detectors on
random subspaces, in: H. Kitagawa, Y. Ishikawa, Q. Li, C. Watanabe (Eds.), Database Systems for Advanced
Applications, Springer Berlin Heidelberg, Berlin, Heidelberg, 2010, pp. 368–383.
[125] D. B. Araya, K. Grolinger, H. F. ElYamany, M. A. Capretz, G. Bitsuamlak, An ensemble learning framework for
anomaly detection in building energy consumption, Energy and Buildings 144 (2017) 191 – 206.
[126] Z. Wun, P. Shi, D. Luo, J. Luo, Research on anomaly detection method for electro-data, in: 2019 IEEE Sustainable
Power and Energy Conference (iSPEC), 2019, pp. 477–481.
[127] R. Primartha, B. A. Tama, Anomaly detection using random forest: A performance revisited, in: 2017 Interna-
tional Conference on Data and Software Engineering (ICoDSE), 2017, pp. 1–6.
[128] Z. Ouyang, X. Sun, J. Chen, D. Yue, T. Zhang, Multi-view stacking ensemble for power consumption anomaly
detection in the context of industrial internet of things, IEEE Access 6 (2018) 9623–9631.
[129] A. Lazarevic, V. Kumar, Feature bagging for outlier detection, in: Proceedings of the Eleventh ACM SIGKDD
International Conference on Knowledge Discovery in Data Mining, KDD ’05, Association for Computing
Machinery, New York, NY, USA, 2005, p. 157–166.
[130] D. B. Araya, K. Grolinger, H. F. ElYamany, M. A. M. Capretz, G. Bitsuamlak, Collective contextual anomaly
detection framework for smart buildings, in: 2016 International Joint Conference on Neural Networks (IJCNN),
2016, pp. 511–518.
[131] X. Gu, L. Akoglu, A. Rinaldo, Statistical analysis of nearest neighbor methods for anomaly detection, in: The
33rd Conference on Neural Information Processing Systems (NeurIPS 2019), 2019, pp. 1–17.
[132] C. Zhang, F. Wang, Multi-feature fusion based anomaly electro-data detection in smart grid, in: 2018 15th
International Symposium on Pervasive Systems, Algorithms and Networks (I-SPAN), 2018, pp. 54–59.

35
 A PREPRINT - F EBRUARY 15, 2021

[133] T. Yijia, G. Hang, Anomaly detection of power consumption based on waveform feature recognition, in: 2016
11th International Conference on Computer Science Education (ICCSE), 2016, pp. 587–591.
[134] L. Tran, L. Fan, C. Shahabi, Distance-based outlier detection in data streams, Proc. VLDB Endow. 9 (12) (2016)
1089–1100.
[135] W. Huo, W. Wang, W. Li, Anomalydetect: An online distance-based anomaly detection algorithm, in: J. Miller,
E. Stroulia, K. Lee, L.-J. Zhang (Eds.), Web Services – ICWS 2019, Springer International Publishing, Cham,
2019, pp. 63–79.
[136] H. Fan, O. R. Zaïane, A. Foss, J. Wu, Resolution-based outlier factor: Detecting the top-n most outlying data
points in engineering data, Knowl. Inf. Syst. 19 (1) (2009) 31–51.
[137] W. Mao, X. Cao, Q. zhou, T. Yan, Y. Zhang, Anomaly detection for power consumption data based on isolated
forest, in: 2018 International Conference on Power System Technology (POWERCON), 2018, pp. 4169–4174.
[138] F. T. Liu, K. M. Ting, Z. Zhou, Isolation forest, in: 2008 Eighth IEEE International Conference on Data Mining,
2008, pp. 413–422.
[139] S. Lee, H. K. Kim, Adsas: Comprehensive real-time anomaly detection system, in: B. B. Kang, J. Jang (Eds.),
Information Security Applications, Springer International Publishing, Cham, 2019, pp. 29–41.
[140] J. Gao, X. Song, Q. Wen, P. Wang, L. Sun, H. Xu, Robusttad: Robust time series anomaly detection via
decomposition and convolutional neural networks (2020). arXiv:2002.09545.
[141] H. Qiu, Y. Tu, Y. Zhang, Anomaly detection for power consumption patterns in electricity early warning system,
in: 2018 Tenth International Conference on Advanced Computational Intelligence (ICACI), 2018, pp. 867–873.
[142] M. Petladwala, Y. Ishii, M. Sendoda, R. Kondo, Canonical correlation based feature extraction with application to
anomaly detection in electric appliances, in: ICASSP 2019 - 2019 IEEE International Conference on Acoustics,
Speech and Signal Processing (ICASSP), 2019, pp. 2737–2741.
[143] T. Andrysiak, u. Saganowski, P. Kiedrowski, Anomaly detection in smart metering infrastructure with the use of
time series analysis, Journal of Sensors 2017 (2017) 1–15.
[144] Z. Ouyang, X. Sun, D. Yue, Hierarchical time series feature extraction for power consumption anomaly detection,
in: K. Li, Y. Xue, S. Cui, Q. Niu, Z. Yang, P. Luk (Eds.), Advanced Computational Methods in Energy, Power,
Electric Vehicles, and Their Integration, Springer Singapore, Singapore, 2017, pp. 267–275.
[145] O. Zyabkina, M. Domagk, J. Meyer, P. Schegner, A feature-based method for automatic anomaly identification
in power quality measurements, in: 2018 IEEE International Conference on Probabilistic Methods Applied to
Power Systems (PMAPS), 2018, pp. 1–6.
[146] L. Mookiah, C. Dean, W. Eberle, Graph-based anomaly detection on smart grid data, in: The Thirtieth Interna-
tional Flairs Conference, 2017, pp. 1–6.
[147] P. Lipcak, M. Macak, B. Rossi, Big data platform for smart grids power consumption anomaly detection, in:
2019 Federated Conference on Computer Science and Information Systems (FedCSIS), 2019, pp. 771–780.
[148] S. W. Yen], S. Morris, M. A. Ezra, T. J. Huat], Effect of smart meter data collection frequency in an early
detection of shorter-duration voltage anomalies in smart grids, International Journal of Electrical Power & Energy
Systems 109 (2019) 1 – 8.
[149] S.-C. Yip, W.-N. Tan, C. Tan, M.-T. Gan, K. Wong, An anomaly detection framework for identifying energy
theft and defective meters in smart grids, International Journal of Electrical Power & Energy Systems 101 (2018)
189 – 203.
[150] M. Peña, F. Biscarri, J. I. Guerrero, I. Monedero, C. León, Rule-based system to detect energy efficiency
anomalies in smart buildings, a data mining approach, Expert Systems with Applications 56 (2016) 242 – 255.
[151] S. Jain, K. A. Choksi, N. M. Pindoriya, Rule-based classification of energy theft and anomalies in consumers
load demand profile, IET Smart Grid 2 (4) (2019) 612–624.
[152] O. Linda, D. Wijayasekara, M. Manic, C. Rieger, Computational intelligence based anomaly detection for
building energy management systems, in: 2012 5th International Symposium on Resilient Control Systems,
2012, pp. 77–82.
[153] C. Chen, D. Cook, Energy outlier detection in smart environments, in: Artificial Intelligence and Smarter Living,
Vol. WS-11-07 of AAAI Workshops, AAAI, 2011.
[154] M. M. Breunig, H.-P. Kriegel, R. T. Ng, J. Sander, Lof: Identifying density-based local outliers, SIGMOD Rec.
29 (2) (2000) 93–104.

36
 A PREPRINT - F EBRUARY 15, 2021

[155] Z. He, X. Xu, S. Deng, Discovering cluster-based local outliers, Pattern Recognition Letters 24 (9) (2003) 1641 –
1650.
[156] F. Giannoni, M. Mancini, F. Marinelli, Anomaly detection models for iot time series data (2018). arXiv:
1812.00890.
[157] Y. Zhou, W. Hu, Y. Min, L. Zheng, B. Liu, R. Yu, Y. Dong, A semi-supervised anomaly detection method for
wind farm power data preprocessing, in: 2017 IEEE Power Energy Society General Meeting, 2017, pp. 1–5.
[158] S. Kim, N. W. Cho, Y. J. Lee, S.-H. Kang, T. Kim, H. Hwang, D. Mun, Application of density-based outlier
detection to database activity monitoring, Information Systems Frontiers 15 (1) (2013) 55–65.
[159] J. Hou, B. Xiao, A data-driven clustering approach for fault diagnosis, IEEE Access 5 (2017) 26512–26520.
[160] M. Ahmed, A. N. Mahmood, M. R. Islam, A survey of anomaly detection techniques in financial domain, Future
Generation Computer Systems 55 (2016) 278–288.
[161] L. Mookiah, C. Dean, W. Eberle, Graph-based anomaly detection on smart grid data, in: Proceedings of the
Thirtieth International Florida Artificial Intelligence Research Society Conference, 2017, pp. 306–311.
[162] L. Akoglu, H. Tong, D. Koutra, Graph based anomaly detection and description: A survey, Data Min. Knowl.
Discov. 29 (3) (2015) 626–688.
[163] M. Davis, W. Liu, P. Miller, G. Redpath, Detecting anomalies in graphs with numeric labels, Association for
Computing Machinery, New York, NY, USA, 2011.
[164] A. Rahmani, S. Afra, O. Zarour, O. Addam, N. Koochakzadeh, K. Kianmehr, R. Alhajj, J. Rokne, Graph-
based approach for outlier detection in sequential data and its application on stock market and weather data,
Knowledge-Based Systems 61 (2014) 89 – 97.
[165] A. Farag, H. Abdelkader, R. Salem, Parallel graph-based anomaly detection technique for sequential data, Journal
of King Saud University-Computer and Information Sciences (2019) 1–9.doi:10.1016/j.jksuci.2019.09.
009.
[166] L. Ruff, R. A. Vandermeulen, N. Görnitz, A. Binder, E. Müller, K.-R. Müller, M. Kloft, Deep semi-supervised
anomaly detection (2019). arXiv:1906.02694.
[167] C. Fan, F. Xiao, Y. Zhao, J. Wang, Analytical investigation of autoencoder-based methods for unsupervised
anomaly detection in building energy data, Applied Energy 211 (2018) 1123 – 1135.
[168] X. Wang, I. Yang, S. Ahn, Sample efficient home power anomaly detection in real time using semi-supervised
learning, IEEE Access 7 (2019) 139712–139725.
[169] H. Song, Z. Jiang, A. Men, B. Yang, A hybrid semi-supervised anomaly detection model for high-dimensional
data, Computational Intelligence and Neuroscience 2017 (2017) 1–9.
[170] H. Janetzko, F. Stoffel, S. Mittelstädt, D. A. Keim, Anomaly detection for visual analytics of power consumption
data, Computers & Graphics 38 (2014) 27 – 37.
[171] N. Cao, C. Lin, Q. Zhu, Y. Lin, X. Teng, X. Wen, Voila: Visual anomaly detection and monitoring with streaming
spatiotemporal data, IEEE Transactions on Visualization and Computer Graphics 24 (1) (2018) 23–33.
[172] J.-S. Chou, A. S. Telaga, W. K. Chong, G. E. Gibson, Early-warning application for real-time detection of energy
consumption anomalies in buildings, Journal of Cleaner Production 149 (2017) 711 – 722.
[173] A. Capozzoli, M. S. Piscitelli, S. Brandi, D. Grassi, G. Chicco, Automated load pattern learning and anomaly
detection for enhancing energy management in smart buildings, Energy 157 (2018) 336 – 352.
[174] V. Saragadam, J. Wang, X. Li, A. C. Sankaranarayanan, Compressive spectral anomaly detection, in: 2017 IEEE
International Conference on Computational Photography (ICCP), 2017, pp. 1–9.
[175] Y. Xia, Z. Zhao, H. Zhang, Distributed anomaly event detection in wireless networks using compressed sensing,
in: 2011 11th International Symposium on Communications Information Technologies (ISCIT), 2011, pp.
250–255.
[176] W. Wang, D. Wang, S. Jiang, S. Qin, L. Xue, Anomaly detection in big data with separable compressive
sensing, in: Q. Liang, J. Mu, W. Wang, B. Zhang (Eds.), Proceedings of the 2015 International Conference on
Communications, Signal Processing, and Systems, Springer Berlin Heidelberg, Berlin, Heidelberg, 2016, pp.
589–594.
[177] M. Levorato, U. Mitra, Fast anomaly detection in smartgrids via sparse approximation theory, in: 2012 IEEE 7th
Sensor Array and Multichannel Signal Processing Workshop (SAM), 2012, pp. 5–8.

37
 A PREPRINT - F EBRUARY 15, 2021

[178] C. Liu, S. Ghosal, Z. Jiang, S. Sarkar, An unsupervised anomaly detection approach using energy-based
spatiotemporal graphical modeling, Cyber-Physical Systems 3 (1-4) (2017) 66–102.
[179] A. Munawar, P. Vinayavekhin, G. De Magistris, Spatio-temporal anomaly detection for industrial robots through
prediction in unsupervised feature space, in: 2017 IEEE Winter Conference on Applications of Computer Vision
(WACV), 2017, pp. 1017–1025.
[180] Z. Yang, N. Japkowicz, Anomaly behaviour detection based on meta-morisita index for large scale spatio-temporal
data set, Journal of Big Data 5 (2018) 1–28.
[181] H. H. Bosman, G. Iacca, A. Tejada, H. J. Wörtche, A. Liotta, Spatial anomaly detection in sensor networks using
neighborhood information, Information Fusion 33 (2017) 41 – 56.
[182] Enetics, Available online: https://www.enetics.com/Products/Software/
Non-Intrusive-Load-Monitoring-NILM, accessed: 2020-06-23.
[183] M. Dilraj, K. Nimmy, S. Sankaran, Towards behavioral profiling based anomaly detection for smart homes, in:
TENCON 2019 - 2019 IEEE Region 10 Conference (TENCON), 2019, pp. 1258–1263.
[184] H. Rashid, N. Batra, P. Singh, Rimor: Towards identifying anomalous appliances in buildings, BuildSys ’18,
Association for Computing Machinery, New York, NY, USA, 2018, p. 33–42.
[185] R. B. Kabler, R. L. Lutes, A. C. Briançon, C. S. Crawford, C. A. Giacoponello, J. F. Johnson, V. A. Jara-Olivares,
M. A. Epard, S. J. Goldberg, J. B. Lancaster, Monitoring and fault detection of electrical appliances for ambient
intelligence, in: United States Patent, no. US9476935B2, 2014, pp. 1–24.
[186] C. Sardianos, I. Varlamis, C. Chronis, G. Dimitrakopoulos, Y. Himeur, A. Alsalemi, F. Bensaali, A. Amira, A
model for predicting room occupancy based on motion sensor data, in: 2020 IEEE International Conference on
Informatics, IoT, and Enabling Technologies (ICIoT), 2020, pp. 394–399.
[187] Y. Laaroussi, M. Bahrar, M. El Mankibi, A. Draoui, A. Si-Larbi, Occupant presence and behavior: A major issue
for building energy performance simulation and assessment, Sustainable Cities and Society 63 (2020) 102420.
[188] W. Kleiminger, C. Beckel, T. Staake, S. Santini, Occupancy detection from electricity consumption data,
BuildSys’13, Association for Computing Machinery, New York, NY, USA, 2013, p. 1–8.
[189] A. Akbar, M. Nati, F. Carrez, K. Moessner, Contextual occupancy detection for smart office by pattern recognition
of electricity consumption data, in: 2015 IEEE International Conference on Communications (ICC), 2015, pp.
561–566.
[190] Y. Gao, A. Schay, D. Hou, Occupancy detection in smart housing using both aggregated and appliance-specific
power consumption data, in: 2018 17th IEEE International Conference on Machine Learning and Applications
(ICMLA), 2018, pp. 1296–1303.
[191] G. Violatto, A. Pandharipande, S. Li, L. Schenato, Classification of occupancy sensor anomalies in connected
indoor lighting systems, IEEE Internet of Things Journal 6 (4) (2019) 7175–7182.
[192] M. Nabil, M. Ismail, M. Mahmoud, M. Shahin, K. Qaraqe, E. Serpedin, Deep recurrent electricity theft detection
in ami networks with random tuning of hyper-parameters, in: 2018 24th International Conference on Pattern
Recognition (ICPR), 2018, pp. 740–745.
[193] V. B. Krishna, K. Lee, G. A. Weaver, R. K. Iyer, W. H. Sanders, F-deta: A framework for detecting electricity
theft attacks in smart grids, in: 2016 46th Annual IEEE/IFIP International Conference on Dependable Systems
and Networks (DSN), 2016, pp. 407–418.
[194] P. Visconti, P. Costantini, R. de Fazio, A. Lay-Ekuakille, L. Patrono, A sensors-based monitoring system of
electrical consumptions and home parameters remotely managed by mobile app for elderly habits’ control, in:
2019 IEEE 8th International Workshop on Advances in Sensors and Interfaces (IWASI), 2019, pp. 264–269.
[195] L. Patrono, P. Primiceri, P. Rametta, I. Sergi, P. Visconti, An innovative approach for monitoring elderly behavior
by detecting home appliance’s usage, in: 2017 25th International Conference on Software, Telecommunications
and Computer Networks (SoftCOM), 2017, pp. 1–7.
[196] L. Patrono, P. Rametta, J. Meis, Unobtrusive detection of home appliance’s usage for elderly monitoring, in:
2018 3rd International Conference on Smart and Sustainable Technologies (SpliTech), 2018, pp. 1–6.
[197] Y. Himeur, A. Elsalemi, F. Bensaali, A. Amira, The emergence of hybrid edge-cloud computing for energy
efficiency in buildings, in: Proceedings of SAI Intelligent Systems Conference, 2021, pp. 1–12.
[198] C. Zhang, W. Ji, Edge computing enabled production anomalies detection and energy-efficient production
decision approach for discrete manufacturing workshops, IEEE Access (2020) 1–11.

38
 A PREPRINT - F EBRUARY 15, 2021

[199] Y. Luo, W. Li, S. Qiu, Anomaly detection based latency-aware energy consumption optimization for iot data-flow
services, Sensors 20 (1) (2020) 1–20.
[200] M. Scarpiniti, E. Baccarelli, A. Momenzadeh, A. Uncini, Smartfog: Training the fog for the energy-saving
analytics of smart-meter data, Applied Sciences 9 (19) (2019) 1–14.
[201] Y. Liu, Z. Pang, M. Karlsson, S. Gong, Anomaly detection based on machine learning in iot-based vertical plant
wall for indoor climate control, Building and Environment 183 (2020) 107212.
[202] J. Zhang, Z. Zhou, S. Li, L. Gan, X. Zhang, L. Qi, X. Xu, W. Dou, Hybrid computation offloading for smart
home automation in mobile cloud computing, Personal and Ubiquitous Computing 22 (2017) 121–134.
[203] A. Anjomshoaa, F. Duarte, D. Rennings, T. J. Matarazzo, P. deSouza, C. Ratti, City scanner: Building and
scheduling a mobile sensing platform for smart city services, IEEE Internet of Things Journal 5 (6) (2018)
4567–4579.
[204] S. Izumi, S. Azuma, Real-time pricing by data fusion on networks, IEEE Transactions on Industrial Informatics
14 (3) (2018) 1175–1185.
[205] Power laws: Detecting anomalies in usage, Available online: https://www.drivendata.org/, accessed:
2020-12-25.
[206] A. S. Uttama Nambi, A. Reyes Lua, V. R. Prasad, Loced: Location-aware energy disaggregation framework,
in: Proceedings of the 2Nd ACM International Conference on Embedded Systems for Energy-Efficient Built
Environments, BuildSys ’15, ACM, New York, NY, USA, 2015, pp. 45–54.
[207] M. Nabil, M. Ismail, M. M. E. A. Mahmoud, W. Alasmary, E. Serpedin, Ppetd: Privacy-preserving electricity
theft detection scheme with load monitoring and billing for ami networks, IEEE Access 7 (2019) 96334–96348.
[208] D. Yao, M. Wen, X. Liang, Z. Fu, K. Zhang, B. Yang, Energy theft detection with energy privacy preservation in
the smart grid, IEEE Internet of Things Journal 6 (5) (2019) 7659–7669.
[209] Y. Himeur, A. Alsalemi, F. Bensaali, A. Amira, Effective non-intrusive load monitoring of buildings based on a
novel multi-descriptor fusion with dimensionality reduction, Applied Energy 279 (2020) 115872.
[210] S. A. Salinas, P. Li, Privacy-preserving energy theft detection in microgrids: A state estimation approach, IEEE
Transactions on Power Systems 31 (2) (2016) 883–894.
[211] Y. Himeur, A. Alsalemi, F. Bensaali, A. Amira, Building power consumption datasets: Survey, taxonomy and
future directions, Energy and Buildings 227 (2020) 110404.
[212] A. Liu, G. Zhang, J. Lu, Concept drift detection based on anomaly analysis, in: C. K. Loo, K. S. Yap, K. W.
Wong, A. Teoh, K. Huang (Eds.), Neural Information Processing, Springer International Publishing, Cham, 2014,
pp. 263–270.
[213] H. Tian, N. L. D. Khoa, A. Anaissi, Y. Wang, F. Chen, Concept drift adaption for online anomaly detection in
structural health monitoring, in: Proceedings of the 28th ACM International Conference on Information and
Knowledge Management, CIKM ’19, Association for Computing Machinery, New York, NY, USA, 2019, p.
2813–2821.
[214] X. Xie, Z. Jin, J. Wang, L. Yang, Y. Lu, T. Li, Confidence guided anomaly detection model for anti-concept drift
in dynamic logs, Journal of Network and Computer Applications 162 (2020) 102659.
[215] H. Akhavan-Hejazi, H. Mohsenian-Rad, Power systems big data analytics: An assessment of paradigm shift
barriers and prospects, Energy Reports 4 (2018) 91 – 100.
[216] Y. Himeur, A. Alsalemi, F. Bensaali, A. Amira, I. Varlamis, G. Bravos, C. Sardianos, Dimitrakopoulos, Techno-
economic analysis of building energy efficiency systems based on behavioral change: A case study of a novel
micro-moments based solution, Applied Energy (2020) 1–18.
[217] Ecoisme, Available online: https://ecoisme.com, accessed: 2020-12-21.
[218] Informetis, Available online: https://www.informetis.com/en/, accessed: 2020-12-23.
[219] Verv Energy, Available online: https://verv.energy/, accessed: 2020-12-23.
[220] Neurio, Neurio sensor W1TM overview, Product specification, Neurio Technology Inc; Vancouver, BC, Canada
(2019).
[221] WiBeee, Available online: https://wibeee.com/en/, accessed: 2020-12-23.
[222] Smart Impulse, Available online: https://www.smart-impulse.com/en/nilm-technology/, accessed:
2020-12-25.
[223] Verdigris, Available online: https://verdigris.co/, accessed: 2020-12-27.

39
 A PREPRINT - F EBRUARY 15, 2021

[224] Voltaware, Available online: https://voltaware.com/support/my-disaggregation, accessed: 2020-12-

28.
[225] HOMEpulse: Creating value from energy data: the disaggregation challenge, Available online: https://www.
homepulse.fr/, accessed: 2020-12-29.
[226] Hive Starter Pack, Available online: https://www.hivehome.com, accessed: 2020-12-21.
[227] DiG, your energy consultant, Available online: https://intelen.com/us/, accessed: 2020-12-27.
[228] Real-time Energy Monitoring Software, Available online: https://harksys.com/, accessed: 2020-12-28.
[229] Enertalk: Energy Monitoring with Big Data Analysis, Available online: https://www.enertalk.com/home/,
accessed: 2020-12-28.
[230] M. Andoni, V. Robu, D. Flynn, S. Abram, D. Geach, D. Jenkins, P. McCallum, A. Peacock, Blockchain
technology in the energy sector: A systematic review of challenges and opportunities, Renewable and Sustainable
Energy Reviews 100 (2019) 143 – 174.
[231] Y. Himeur, A. Elsalemi, F. Bensaali, A. Amira, An intelligent non-intrusive load monitoring scheme based on
2D phase encoding of power signals, International Journal of Intelligent Systems 36 (1) (2021) 72–93.
[232] Y. Himeur, A. Alsalemi, F. Bensaali, A. Amira, Smart non-intrusive appliance identification using a novel local
power histogramming descriptor with an improved k-nearest neighbors classifier, Sustainable Cities and Society
67 (2021) 102764.
[233] H. Rashid, P. Singh, Energy disaggregation for identifying anomalous appliance, in: Proceedings of the 4th
ACM International Conference on Systems for Energy-Efficient Built Environments, Association for Computing
Machinery, 2017.
[234] Y. Himeur, A. Elsalemi, F. Bensaali, A. Amira, Appliance identification using a histogram post-processing of 2d
local binary patterns for smart grid applications, in: Proc. 25th International Conference on Pattern Recognition
(ICPR), 2020, pp. 1–8.
[235] K. Amarasinghe, K. Kenney, M. Manic, Toward explainable deep neural network based anomaly detection, in:
2018 11th International Conference on Human System Interaction (HSI), 2018, pp. 311–317.
[236] J. Kauffmann, L. Ruff, G. Montavon, K.-R. Müller, The clever hans effect in anomaly detection (2020).
arXiv:2006.10609.
[237] L. Antwarg, R. M. Miller, B. Shapira, L. Rokach, Explaining anomalies detected by autoencoders using shap
(2019). arXiv:1903.02407.
[238] P. Liznerski, L. Ruff, R. A. Vandermeulen, B. J. Franks, M. Kloft, K.-R. Müller, Explainable deep one-class
classification (2020). arXiv:2007.01760.
[239] C. Sardianos, C. Chronis, I. Varlamis, G. Dimitrakopoulos, Y. Himeur, A. Alsalemi, F. Bensaali, A. Amira,
Real-time personalised energy saving recommendations, in: 2020 International Conferences on Internet of Things
(iThings) and IEEE Green Computing and Communications (GreenCom) and IEEE Cyber, Physical and Social
Computing (CPSCom) and IEEE Smart Data (SmartData) and IEEE Congress on Cybermatics (Cybermatics),
IEEE, 2020, pp. 366–371.
[240] S. K. Bose, B. Kar, M. Roy, P. K. Gopalakrishnan, A. Basu, Adepos: Anomaly detection based power saving for
predictive maintenance using edge computing, ASPDAC ’19, Association for Computing Machinery, New York,
NY, USA, 2019, p. 597–602.
[241] A. Sayed, A. Elsalemi, Y. Himeur, F. Bensaali, A. Amira, Endorsing energy efficiency through accurate appliance-
level power monitoring, automation and data visualization, in: The 4th International Conference on Networking,
Information Systems & Security (NISS 2021), 2021, pp. 1–13.
[242] A. Alsalemi, Y. Himeur, F. Bensaali, A. Amira, C. Sardianos, I. Varlamis, G. Dimitrakopoulos, Achieving
domestic energy efficiency using micro-moments and intelligent recommendations, IEEE Access 8 (2020)
15047–15055.
[243] S. Aberkane, M. Elarbi, Deep reinforcement learning for real-world anomaly detection in surveillance videos, in:
2019 6th International Conference on Image and Signal Processing and their Applications (ISPA), 2019, pp. 1–5.
[244] S. M. A. Shabestary, B. Abdulhai, Deep learning vs. discrete reinforcement learning for adaptive traffic signal
control, in: 2018 21st International Conference on Intelligent Transportation Systems (ITSC), 2018, pp. 286–293.
[245] M. Gregurić, M. Vujić, C. Alexopoulos, M. Miletić, Application of deep reinforcement learning in traffic signal
control: An overview and impact of open traffic data, Applied Sciences 10 (11) (2020) 1–25.

40
 A PREPRINT - F EBRUARY 15, 2021

[246] N. C. Luong, D. T. Hoang, S. Gong, D. Niyato, P. Wang, Y. Liang, D. I. Kim, Applications of deep reinforcement
learning in communications and networking: A survey, IEEE Communications Surveys Tutorials 21 (4) (2019)
3133–3174.
[247] T. Liu, Z. Tan, C. Xu, H. Chen, Z. Li, Study on deep reinforcement learning techniques for building energy
consumption forecasting, Energy and Buildings 208 (2020) 109675.
[248] J. Zhao, N. Cao, Z. Wen, Y. Song, Y. Lin, C. Collins, #fluxflow: Visual analysis of anomalous information
spreading on social media, IEEE Transactions on Visualization and Computer Graphics 20 (12) (2014) 1773–
1782.
[249] Y. Shi, Y. Liu, H. Tong, J. He, G. Yan, N. Cao, Visual analytics of anomalous user behaviors: A survey (2019).
arXiv:1905.06720.
[250] S. Guo, Z. Jin, Q. Chen, D. Gotz, H. Zha, N. Cao, Visual anomaly detection in event sequence data, in: 2019
IEEE International Conference on Big Data (Big Data), 2019, pp. 1125–1130.
[251] A. Al-Kababji, A. Alsalemi, Y. Himeur, R. fernandez, F. Bensaali, A. Amira, N. Fetais, Interactive visual
analytics for residential energy big data, Information Visualization (2020) 1–18.
[252] A. Al-Kababji, A. Alsalemi, Y. Himeur, F. Bensaali, A. Amira, R. Fernandez, N. Fetais, Energy data visualizations
on smartphones for triggering behavioral change: Novel vs. conventional, in: 2020 2nd Global Power, Energy
and Communication Conference (GPECOM), 2020, pp. 312–317.
[253] N. Batra, R. Kukunuri, A. Pandey, R. Malakar, R. Kumar, O. Krystalakos, M. Zhong, P. Meira, O. Parson, Towards
reproducible state-of-the-art energy disaggregation, BuildSys ’19, Association for Computing Machinery, New
York, NY, USA, 2019, p. 193–202.
[254] N. Batra, J. Kelly, O. Parson, H. Dutta, W. Knottenbelt, A. Rogers, A. Singh, M. Srivastava, Nilmtk: An open
source toolkit for non-intrusive load monitoring, in: Proceedings of the 5th International Conference on Future
Energy Systems, e-Energy ’14, Association for Computing Machinery, New York, NY, USA, 2014, p. 265–276.
[255] G. A. Kaissis, M. R. Makowski, D. Rückert, R. F. Braren, Secure, privacy-preserving and federated machine
learning in medical imaging, Nature Machine Intelligence 2 (2020) 305–311.
[256] M. Hao, H. Li, G. Xu, S. Liu, H. Yang, Towards efficient and privacy-preserving federated deep learning, in:
ICC 2019 - 2019 IEEE International Conference on Communications (ICC), 2019, pp. 1–6.
[257] R. Madurai Elavarasan, G. Shafiullah, R. Kannadasan, V. Mudgal, M. Arif, T. Jamal, S. Senthilkumar, V. Sri-
RajaBalaguru, K. Reddy, U. Subramaniam, Covid-19: Impact analysis and recommendations for power sector
operation, Applied Energy (2020) 115739.
[258] C. Sardianos, C. Chronis, I. Varlamis, G. Dimitrakopoulos, Y. Himeur, A. Alsalemi, F. Bensaali, A. Amira,
Smart fusion of sensor data and human feedback for personalised energy-saving recommendations, International
Journal of Intelligent Systems (2021) 1–20.
[259] M. Nilashi, S. Asadi, R. A. Abumalloh, S. Samad, O. Ibrahim, Intelligent recommender systems in the covid-19
outbreak: The case of wearable healthcare devices, Journal of Soft Computing and Decision Support Systems
7 (4) (2020) 8–12.
[260] J. Budd, B. Miller, E. Manning, V. Lampos, M. Zhuang, M. Edelstein, G. Rees, V. Emery, M. Stevens, N. Keegan,
M. Short, D. Pillay, E. Manley, I. Cox, D. Heymann, A. Johnson, R. McKendry, Digital technologies in the
public-health response to covid-19, Nature Medicine 26 (2020) 1–10.
[261] A. Alsalemi, Y. Himeur, F. Bensaali, A. Amira, C. Sardianos, C. Chronis, I. Varlamis, G. Dimitrakopoulos, A
micro-moment system for domestic energy efficiency analysis, IEEE Systems Journal (2020) 1–8.
[262] Y. Himeur, A. Alsalemi, F. Bensaali, A. Amira, , C. Sardianos, Dimitrakopoulos, I. Varlamis, A survey of
recommender systems for energy efficiency in buildings: Principles, challenges and prospects, Information
Fusion (2020) 1–33.
[263] Y. Zhang, X. Chen, Explainable recommendation: A survey and new perspectives, Foundations and Trends® in
Information Retrieval 14 (1) (2020) 1–101.
[264] C. Sardianos, I. Varlamis, C. Chronis, G. Dimitrakopoulos, Y. Himeur, A. Alsalemi, F. Bensaali, A. Amira, Data
analytics, automations, and micro-moment based recommendations for energy efficiency, in: 2020 IEEE Sixth
International Conference on Big Data Computing Service and Applications (BigDataService), IEEE, 2020, pp.
96–103.

41