Online Prediction and Learning
Graduate Course, IISc, 2022
Lecture notes consists of materials from my coursework at IISc.
- Lec-1: Introduction to Online Learning.
- Lec-2: Predictions with Experts
- Lec-3: Halving Algorithm
- Lec-4: Weighted Majority (WMAJ) & Random-WMAJ Algorithms
- Lec-5: Convex Analysis and Online Convex Learning
- Lec-6: Follow The Leader (FTL)
- Lec-7: Introduction to Follow the Regularized Leader (FTRL)
- Lec-8: Analysis of FTRL
- Lec-9: FTRL futher examples and analysis
- Lec-10: Geometrix view of FTRL
- Lec-11: Convex Geometry & Dual Functions
- Lec-12: Frenchel Duality
- Lec-13: Online Mirror Descent
- Lec-14: Optimization Theory view
- Lec-15: Online Convex Optimization with Bandits
- Lec-16: Online Linear Optimization with Bandits Feedback
- Lec-17: Stochastic Bandit Algorithms
- Lec-18: Exploration vs. Exploitation
- Lec-19: Hoeffding’s Inequality and Intro to UCB
- Lec-20: UCB: Regret bound analysis
- Lec-21: Introduction to Thompson Sampling
- Lec-22 & Lec-23: Regret Analysis: Thompson Sampling
- Lec-24: General Stochastic Multiarm Bandits
- Lec-25 & Lec-26 & Lec-27: Fundamental limits of Bandit learning
- Lec-28: Structured Bandits