=================================
Distribution Dependent PAC bounds
=================================
- John Langford

I'll present the intuition behind new techniques that substantially
tighten the typical machine learning sample complexity bounds and
present results comparing various bounds on problems from the UCI
Machine Learning database.