Lectures and reading

Lecture 1+2 Introduction to learning theory, learning from expert advice and the online learning model

Examples of machine learning problems,  classification of learning models, learning from expert advice, the weighted majority + RWM algorithms + their analysis, discussion of lower bounds,

In lecture 2, we covered the Hedge algorithm, and now a naive application to many problems of interest is inefficient. Discussion of online routing. Motivate mathematical optimization

Reading:

• Chapter 1  in book 1

Last year's scribe notes:

• Lecture 1

• Lecture 2

Lecture 3 First-Order Algorithms for Online Convex Optimization

Introduction to convex analysis and mathematical optimization, the online gradient descent and its regret analysis. We derived online shortest paths. 

Reading:

• Chapter 3  in book 1

Lecture 4 Introduction to (Offline) Convex Optimization

Introduction to convex analysis and mathematical optimization and the (projected) gradient descent algorithm, with reductions to non-smooth and non-strongly convex functions.

Reading:

• Chapter 2  in book 1