Deep Learning Seminar

"Rademacher Complexities and Generalization Bounds"

Reisenhofer, Rafael

In this talk, I will review basic concepts and proofs regarding the notion of Rademacher complexity. In particular, we will see how the Rademacher complexity of a hypothesis class can be used to bound the generelization error of a learning task and derive an upper bound for the Rademacher complexity of a linear classifier.

« back