Machine Learning Courseworks (Imperial)

This repository contains solutions for different Machine Learning courseworks at Imperial College London (2022-2024).

COMP70050: Introduction-to-Machine-Learning CW 1 (2022-2023) - Decision Tree Coursework

• Built a decision tree for WiFi localization with pruning and cross-validation.

📜 Coursework 1 Specification
📑 Coursework 1 Report
📂 Coursework 1 Files

COMP70050: Introduction-to-Machine-Learning CW 2 (2022-2023) - Artificial Neural Networks

• Developed a modular neural network library from first principles using NumPy.

📜 Coursework 2 Specification
📑 Coursework 2 Report
📂 Coursework 2 Files

COMP60006: Computer Vision CW 1 (2022-2023)

• Implemented image filters from scratch, including moving average, Gaussian smoothing, and Sobel filters, to reduce noise and enhance image quality.
• Applied edge detection techniques to identify important features, experimenting with different filter sizes and methods for better accuracy.
• Transitioned to PyTorch’s Conv2D filtering for a more efficient and optimized approach to image processing.

📜 Coursework 1 Specification
📓 Coursework 1 Notebook | 🔗 View in nbviewer | 📝 View PDF

COMP60006: Computer Vision CW 2 (2022-2023)

• Trained and evaluated a U-Net model for brain tumor segmentation on 2D MRI slices.

📜 Coursework 2 Specification
📓 Coursework 2 Notebook | 🔗 View in nbviewer | 📝 View PDF

COMP70015: Mathematics for Machine Learning CW (2023-2024)

• Used gradient descent to optimize mathematical functions and study how different types behave during convergence.
• Explored how learning rates, function properties, and stopping criteria affect optimization, using tools like finite differences, autograd, and visualizations.

📜 Coursework Specification
📓 Coursework Notebook | 🔗 View in nbviewer

MATH70134: Mathematical Foundations of Machine Learning CW 1 (2023-2024)

• Explored key theoretical concepts in machine learning.
• Implemented and trained a fully connected neural network.
• Linked deep neural networks to Gaussian processes by implementing an NNGP model.

📜 Coursework 1 Specification
📓 Coursework 1 Notebook | 🔗 View in nbviewer

MATH70134: Mathematical Foundations of Machine Learning CW 2 (2023-2024)

• Optimized a diffusion model for generative tasks using a U-Net-based denoising process.
• Built a DeepDream pipeline with a pre-trained Inception-V3 for artistic feature visualizations.

📓 Coursework 2 Notebook | 🔗 View in nbviewer