Drifting Games, Boosting and Online Learning

Drifting games provide a new and useful framework for analyzing learning algorithms. In this talk I will present the framework and show how it is used to derive a new boosting algorithm, called RobustBoost and a new online prediction algorithm, called NormalHedge. I will present two sets of experiments using these algorithms on synthetic and real world data. The first set demonstrates that RobustBoost can learn from mislabeled training data. The second demonstrating an application of NormalHedge to the tracking moving objects.