Explore a comprehensive seminar lecture from the Kolmogorov Seminar series that delves into the classical Vapnik-Chervonenkis theorem, examining why a small VC-dimension of hypotheses space leads to minimal error rates when comparing training and test samples. Learn from this in-depth 2.5-hour discussion that continues the legacy of the computational and descriptional complexity seminars established by Kolmogorov in 1979.
Overview
Syllabus
Vereshchagin explains (classical) Vapnik-Chervonenkis theorem 11.3.2024
Taught by
Kolmogorov-Seminar
Related Courses
-
Improving Solovay's Theorem in Computational Complexity - Part 1
-
Improving Solovay's Theorem - Part 2
-
Machine Learning: Shattering and VC Dimension - Lecture 17B
-
The VC Dimension in Machine Learning - Lecture 18
-
Introduction to Probably Approximately Correct (PAC) Learning
-
Computable PAC Learning - Part 2
