This course introduces fundamental concepts and practical skills in machine learning and deep learning. Students begin with classical techniques such as linear models and decision trees, then progress to modern architectures including Convolutional Neural Networks (CNNs), Variational Autoencoders (VAEs), Diffusion Models, and Transformers for sequence modeling.
Lectures are complemented by weekly hands-on tutorials, where students implement methods in Python using PyTorch. The coursework includes four programming assignments covering applications in computer vision, language processing, and generative modeling.
Prerequisites: Introduction to Probability and Introduction to Computer Science.
Additional course details and resources are available on the Moodle course website.
This course offers an in-depth exploration of Graph Neural Networks (GNNs), covering foundational theories and state-of-the-art methodologies. Core topics include Message Passing Neural Networks (MPNNs), spectral approaches, Graph Transformers, dynamic graphs, and the interplay between large language models (LLMs) and graph structures.
Students will work on hands-on programming exercises using PyTorch Geometric (PyG) and NetworkX. The course emphasizes practical research challenges such as oversmoothing and oversquashing, and techniques for modeling heterogeneous and large-scale graphs.
A key component is a research-oriented project, involving literature review, novel development, implementation, and peer presentation.
Prerequisites: Foundational machine learning knowledge and proficiency in Python.
Additional details and course materials are available on the course website.
