2023 NCKU Image Processing Homework Code

Topic Modelling von Tagesschau-Nachrichtenmeldungen mit NMF

Pipeline de PLN do projeto "Mood Hound" (6º DSM - 2023, FATEC Profº Jessen Vidal - SJC)

Classifies customer complaints into 5 service categories using NMF-based topic modeling and supervised learning (Logistic Regression, RF). Achieved ~85% accuracy on real-world ticket data.

This project aims to identify key topics in Quora questions related to popular applications. We'll use Non-Negative Matrix Factorization (NMF) to extract meaningful themes and patterns from this data.

Statistical machine learning techniques

Non-Negative Matrix Factorization (NMF) is a powerful technique used in audio processing for separating mixed audio sources into their constituent parts

A movie recommendation website powered by an unsupervised learning non-negative matrix factorization algorithm. This project provides users with personalized movie recommendations based on their viewing history and preferences.

Customer insight analysis of product purchase behaviour using EDA, text analysis, and thematic modelling.

This project processes customer complaint data using pandas for data manipulation and applies text preprocessing techniques, including lemmatization, to clean and normalize complaint text. The `tqdm` library provides progress bars for efficient tracking of text processing tasks.

📊 Interactive visualisation tool for exploring Non-negative Matrix Factorization (NMF) clustering results

Two Movie recommenders deployed on Flask app. The recommender are build on NMF and Cosine Similarity model, trained on MovieLens 100k, and automatically updated every 12 hour!

The repository contains files (notebooks, data) for the course work of the 2nd course: "Topic modeling for text document analysis".

Topic modeling on Shopee's 1-star reviews to uncover insights and prevalent topics within the reviews.

My MTech project focuses on classifying hand gestures using surface Electromyography (sEMG) signals to enable intuitive control for assistive and rehabilitation technologies. I worked with large-scale EMG datasets (80,000×8 signals per trial) and built a complete ML pipeline for accurate classification of the hand postures.

Implementation of the Non-negative Multiple Matrix Factorization (NMMF) algorithm proposed in Takeuchi et al, 2013 with some modifications. There is a python native version NMMFlexPy and a R wrapper NMMFlexR

Unsupervised Machine Learning on attacking footballers in Europe’s top 5 leagues (24/25), featuring K-Means Clustering, Anomaly Detection, and an NMF-based similarity recommendation system.

Towards Intra-Tumor Heterogeneity Deconvolution from Bulk RNA-seq data @BoevaLab

Unsupervised classification of products based on their text description (NLP) or image (computer vision)