These are my running notes made for the course CS 726 - Advanced Machine Learning @ IIT Bombay. The notes can be considered a superset of the class content.
The chapter numbers are indexed, with the title after it. The main file to read is 0-main.pdf, which has the content.
Summary of what has been done, and what will be done:
- Probabilistic Modeling - Small overview of probability, graphical models and conditional independence, along with their significance.
- Bayesian Networks - Directed Graphical Models for representation of joint distributions, d-separability.
- Markov Random Fields - Undirected Graphical Models, conditional independencies, drawbacks, conversion to and from Bayesian Networks.
- Inference Queries - Types of inference queries, NP-hardness, 3-SAT, Variable Elimination, Junction Trees, Message Passing, Inference with evidence, Exact and Approximate inference, Belief Propogation
- Learning Potentials - Structure & Potential Learning methods, Training with gradient descent [to be continued]