Extra Chapter: Online Learning
This chapter introduces online learning.
-
Extra Chapter: Online Learning - Introduction
In this chapter we explain the differences between online and batch learning, the extended learning protocol in online learning and the strategies to measure performance in online learning.
-
Extra Chapter: Online Learning - Simple Algorithms
In this chapter we introduce the formalization of online learning algorithms and the Follow the Leader (FTL) algorithm.
-
Extra Chapter: Online Learning - Follow the Leader on Online Linear Optimization
In this chapter we introduce Online Linear Optimization (OLO) problems and explain why Follow the Leader (FTL) can fail on these problems.
-
Extra Chapter: Online Learning - Follow the Regularized Leader
In this chapter we introduce Follow the Regularized Leader (FTRL) as a stable alternative to Follow the Leader (FTL).
-
Extra Chapter: Online Learning - Follow the Leader on Online Quadratic Optimization
In this chapter we prove that Follow the Leader (FTL) works for Online Quadratic Optimization (OQO) problems.
-
Extra Chapter: Online Learning - Online Convex Optimization 1
In this chapter we introduce the class of Online Convex Optimization (OCO) problems and derive Online Gradient Descent (OGD) as a suitable learning algorithm for such cases.
-
Extra Chapter: Online Learning - Online Convex Optimization 2
In this chapter we explain the connection between Online Gradient Descent (OGD) and Follow the Regularized Leader (FTRL) via linearization of convex functions and how this implies regret bounds for OGD.