Machine Learning Projects

This repository contains a collection of machine learning and deep learning projects completed as part of coursework, research, and personal interest. Projects are organized by topic and implemented in Jupyter notebooks using libraries such as scikit-learn, TensorFlow, and PyTorch.

Classification

• classification/biomedical-image/pathmnist-transfer-learning/ - Comparative study of transfer learning effectiveness for biomedical image classification using ResNet-18 on PathMNIST dataset, implementing three approaches: training from scratch, feature extraction with frozen pre-trained weights, and full fine-tuning (PyTorch, torchvision, medmnist)

• classification/malware-family/ - Classifying malware into families using API call sequences with Word2Vec and BERT embeddings combined with machine learning and CNN models (scikit-learn, xgboost, transformers, PyTorch)

• classification/mnist/ - Handwritten digit classification using a neural network (TensorFlow)

• classification/wine-quality/ - Classifying wine quality based on physicochemical properties using a feed-forward neural network (PyTorch)

Generative Models

• generative-models/vae/ - Building a variational eutoencoder to generate and visualize fashion MNIST items in a 2D latent space (TensorFlow)

Bioinformatics

• bioinformatics/svm_braf_v600e_prediction/ - Predicting BRAF V600E inhibitors using support vector machine (scikit-learn)

Getting Started

Each project has its own folder and may include a requirements.txt. To run a notebook:

1. Clone this repo
2. Navigate to the desired project folder
3. Install requirements (if provided)
4. Launch Jupyter Notebook