This repository offers a comprehensive and continuously updated collection of machine learning, probabilistic models, and deep learning notes, slides, and demos. Aimed at intermediate to advanced learners and researchers, it provides deep dives into theoretical concepts, mathematical underpinnings, and practical implementations, serving as a rich, self-paced learning resource.

How It Works

The project curates extensive lecture notes, often supplemented with code examples (e.g., PyTorch, MATLAB demos), covering a vast spectrum from foundational mathematics for ML to advanced topics in generative AI, reinforcement learning, and Bayesian non-parametrics. Its approach integrates rigorous mathematical derivations with practical code analysis, exemplified by detailed breakdowns of Transformer architectures or Kalman filters. The inclusion of live online seminars and recorded video tutorials further enhances the learning experience by offering interactive sessions and supplementary visual explanations.

Quick Start & Requirements

• Installation: No direct installation is required for accessing the notes; they are presented as documentation.
• Prerequisites: A solid understanding of linear algebra, calculus, probability, and statistics is essential. Specific sections may require familiarity with Python and PyTorch.
• Resources: Access to the repository content is free. Seminar registration is available via meetup.com/machine-learning-hong-kong/.
• Links: Meetup

Highlighted Details

• Features weekly live Machine Learning classes in Chinese and bi-weekly English research seminars.
• Provides in-depth analysis of Transformer models using PyTorch, covering advanced techniques like K-V Caching and Multi-Head Latent Attention.
• Offers extensive coverage of Bayesian Non-Parametrics (BNP), including Dirichlet Processes (DP) and Indian Buffet Processes (IBP), with associated code demos.
• Includes detailed mathematical treatments of optimization methods like Gradient Descent variants, Duality, and Conjugate Gradients.

Maintenance & Community

The repository is described as "continuously updated," with ongoing efforts to validate and correct notes. The author actively seeks high-quality PhD students for research collaboration, providing an academic contact point (xuyida@hkbu.edu.hk). No specific community forums (like Discord or Slack) are mentioned.

Licensing & Compatibility

No open-source license is specified in the provided text. This lack of explicit licensing information may pose compatibility concerns for commercial use or integration into proprietary projects.

Limitations & Caveats

Some video tutorials were recorded in 2015 and cover only a fraction (~10-20%) of the notes. Certain advanced topics, such as "Completely Random Measure," are noted as early drafts. While the primary language for notes is English, some live sessions are conducted in Mandarin. The absence of a specified license is a significant caveat for adoption.