Machine Learning Methods
Semestr: Summer
Range: 2P+2C
Completion:
Credits: 6
Programme type: Master
Study form: Fulltime
Course language: English
Summary:
Students will get familiar with machine learning methods that go beyond the standard settings taught in basic ML courses. They will learn methods that work well for tabular and structured data domains (e.g. relational databases), including graph neural networks and recent neuro-symbolic techniques. The course will also teach the students some methods for model interpretability, basics of causality, and reinforcement learning.
Keywords:
Course syllabus:
1. Learning from Tabular data
2. Ensembling and boosting
3. Learning from Structured data
4. Graph Neural Networks
5. Neural-Symbolic methods
6. ML Interpretability
7. ML Operations
8. Potential outcomes - Rubin-Neyman causal model, uplift modeling
9. Intro to “Pearl’s” causality
10. A/B tests and multi-armed bandit problems, UCB algorithm.
11. Bayesian bandits (Thompson sampling). Contextual bandits.
12. Markov decision processes
13. Tabular RL: Q-Learning and SARSA
14. Deep RL: Deep Q-learning. Policy gradient.