Week 1
What is Machine Learning - Simplistic View
Automating automation
Getting computers to program themselves
Writing software is the bottleneck
Let the data do the work instead!
What is Machine Learning – Technical view
Machine learning is an application of artificial intelligence
(AI) that provides systems the ability to automatically learn
and improve from experience without being explicitly
programmed.
Machine learning focuses on the development of computer
programs that can access data and use it to learn for themselves.
Machine learning is a method of data analysis that
automates analytical model building. It is a branch of
artificial intelligence based on the idea that machines
should be able to learn and adapt through experience.
ow is different from AI
Artificial Intelligence is the broader concept of
machines being able to carry out tasks in a way that we
would consider “smart”.
Machine Learning is a current application of AI based
around the idea that we should really just be able to
give machines access to data and let them learn for
themselves.
ow did it all Begin….?
credited to Arthur Samuel in 1959
Why Teach them everything….?
Why not let them learn for themselves
○ Far more efficient
○ Far more effective
Vast e- info available
Code them to process and learn from that info
Code them to think like humans and adapt and learn
continuously
Machine Learning – Pure Convenience with
ccuracy
By using algorithms to build models that uncover
connections, patterns, organizations can make better
decisions without human intervention
Learning is used when:
Human expertise does not exist (navigating on Mars),
Humans are unable to explain their expertise /Black-box human
expertise (speech recognition)
Solution changes in time (diagnosis, fraud detection, routing on a
computer network)
Solution needs to be adapted to particular cases (user biometrics,
personalized news reader, movie/book recommendation)
uzz Words
Big Data
Yota byte
Deep Learning
Chatbot
Natural Language Processing
Data Mining
IoT
Smart City
Where it comes from
Machine learning is primarily concerned with the accuracy
and effectiveness of the computer system.
data
mining control theory
statistics
decision theory
information theory machine
learning
cognitive science
databases
psychological models
evolutionary neuroscience
models
What this course is about
This course aims at introducing you to the exciting and
trending world of Machine learning and teach the
fundamentals so that you could use your human learning
mechanisms to learn further (as required) and apply that
knowledge effectively for any future ML applications
We will learn fundamental concept, some popular machine
learning algorithms and their strengths and weaknesses
To reinforce our learning, we will implement these
algorithms, for real world applications, using python.
Introduction and Significance of Machine Learning, Machine Learning
ontents applications
Review of Important Mathematical/Statistical Concepts
Introduction to Machine Learning Techniques
Supervised, Unsupervised and Reinforcement Learning
Supervised Learning Techniques
Regression Learning Problems
Regression Algorithms
Linear Regression
Multivariable Regression
Feature Selection
Supervised Learning Techniques
Classification Problems
Classification Algorithms
Naïve Bayes
KNN
ANN
Dimensionality Reduction
Unsupervised Learning
Clustering Problems
Clustering Algorithms
K Means
Fuzzy C Means
ourse Policies
2 Classes (1.5 hrs ) each week for 16 weeks
45 mins lecture (little theory and more practice)
45 mins pure practical
○ Python
We will spend first few weeks practicing the fundamental
programming constructs in python
Followed by a refresher on statistics (practical)
Finally by ML algorithm implementation to solve ML Problems
shaAllah
Attendance
Discipline
Machine Learning – Do we really need to
udy it?
“A breakthrough in machine learning would be worth
ten Microsofts” (Bill Gates, Chairman, Microsoft)
“Machine learning is the next Internet”
(Tony Tether, Director, DARPA)
Machine learning is the hot new thing”
(John Hennessy, President, Stanford)
“Web rankings today are mostly a matter of machine learning”
(Prabhakar Raghavan, Dir. Research, Yahoo)
“Machine learning is going to result in a real revolution” (Greg
Papadopoulos, CTO, Sun)
“Machine learning is today’s discontinuity”
(Jerry Yang, CEO, Yahoo)
op Machine Learning Application
Security Lifestyle
Data Security Smart Cars
Personal Security Recommendations
Employee Authorization – Online Search
Amazon NPL
Healthcare Financial Applications
Heart Attack identification Trading
Seizure and stroke prediction Fraud Detection
CAD, etc. Marketing Personalization
What is ML in a Nut Shell???
Tens of thousands of machine learning algorithms
Hundreds new every year
Every machine learning algorithm has three
components:
Representation
Evaluation
Optimization
epresentation
Decision trees
Sets of rules / Logic programs
Instances
Graphical models (Bayes/Markov nets)
Neural networks
Support vector machines
Model ensembles
Etc.
valuation
Accuracy
Precision and recall
Squared error
Likelihood
Posterior probability
Cost / Utility
Margin
Entropy
K-L divergence
Etc.
ptimization
Combinatorial optimization
E.g.: Greedy search
Convex optimization
E.g.: Gradient descent
Constrained optimization
E.g.: Linear programming
Machine Learning Techniques
Supervised Learning
Unsupervised learning
Reinforcement
Learning
upervised Learning
A task of inferring a function from labeled
training data.
The training data consist of a set of training
examples.
Each training example is a pair consisting of an
input object and an output value
The algorithm attempts to learn from the training
examples and predict for unseen (test) data
Finds a model that best maps the input to output (data
to target)
○ A function
Uses the inferred model to predict for the test data
upervised Learning
The input consists of feature(s) describing the input
object and the output is the target value
nsupervised Learning
A task to find hidden patterns in un-labbelled data
Unlike supervised learning , there is no correct answers
and there is no teacher
up Vs. Un Sup
e-enforcement Learning
Allows software agents and machines to automatically
determine the ideal behavior within a specific context,
to maximize its performance
Simple reward feedback is required for the agent to
learn its behavior; this is known as the reinforcement
signal.
e-enforcement Learning
Machine Learning Problems/Algorithms
Regression
Predicts a continuous value
Classification
Predicts a discrete class label
Clustering
Finding a structure in a collection of unlabeled data
The process of organizing objects into groups whose
members are similar in some way
egression Problems
The output is a continuous value
Examples
• The price of a house
• The humidity in environment
• The stock price
• The number of people online at a specific hour
lassification Problem
• The output is a discrete class label
• Example
• Will it rain today (Yes/No)
• Which Animal is it (Cat/Dog/Horse)
• Student Performance Category (Below, Average or Above)
• The severity of a tumor (benign/Malignant)
• Outcome of a surgery (Successful/Unsuccessful)
lustering Problems
Clusters
Associations
Examples
Market Segments
Deploying Police at spots with high crime rates for
maximum coverage and effectiveness
dentify Problem Type/Algorithm
Predict the temperature
Which degree a student will opt based on scores and other factors
Identifying a fruit in an image
What should be the speed of a self driven car on this road
Which team will win this match ?
How much score will my team make in the next match
Which complexity levels should we offer a particular course at?
Would it rain tomorrow
Would the project start on time
How much delay is expected in the starting of this project
What would be the girls to boys ratio in the next session of Computer
Arts.
Which products should be discounted near Eid
ow it goes??
Understanding domain, prior knowledge, and goals
Data integration, selection, cleaning,
pre-processing, etc.
Learning models
Interpreting results
Consolidating and deploying discovered knowledge
Loop
ython Introduction
High Level
Open source and community driven
“Batteries Included”
a standard distribution includes many modules
Dynamic typed
Source can be compiled or run just-in-time
simple python program
x = 42
y = 58
z = ‘Intro to Python Programming’
a = x+ y
print( a, z)
ists
compound data type:
]
.3, 4.5]
, "Hello", "there", 9.8]
se len() to get the length of a list
>> names = [“Beenish", “Homa", “Sherein"]
>> len(names)
sts are mutable - some useful methods
> ids = ["9pti", "2plv", "1crn"]
> ids.append("1alm")
append an element
> ids
ti', '2plv', '1crn', '1alm']
>ids.extend(L) remove an element
Extend the list by appending all the items in the given list; equivalent to a[len(a):] = L.
> del ids[0]
> ids
lv', '1crn', '1alm']
sort by default order
> ids.sort()
> ids
lm', '1crn', '2plv'] reverse the elements in a list
> ids.reverse()
> ids
lv', '1crn', '1alm']
> ids.insert(0, "9pti")
insert an element at some
> ids specified position.
ti', '2plv', '1crn', '1alm']
(Slower than .append())
ictionaries
Dictionaries are lookup tables.
They map from a “key” to a “value”.
symbol_to_name = {
"H": "hydrogen",
"He": "helium",
"Li": "lithium",
"C": "carbon",
"O": "oxygen",
"N": "nitrogen"
}
Duplicate keys are not allowed
Duplicate values are just fine
