Inapproximability of VC Dimension and Littlestone's Dimension
arXiv:1705.09517
Abstract
We study the complexity of computing the VC Dimension and Littlestone's Dimension. Given an explicit description of a finite universe and a concept class (a binary matrix whose -th entry is iff element belongs to concept ), both can be computed exactly in quasi-polynomial time (). Assuming the randomized Exponential Time Hypothesis (ETH), we prove nearly matching lower bounds on the running time, that hold even for approximation algorithms.
arXiv commentTo appear in Conference On Learning Theory (COLT) 2017
References in corpus (5)
- Complexity theoretic limitations on learning DNF's
- Mixture Selection, Mechanism Design, and Signaling
- Finding Overlapping Communities in Social Networks: Toward a Rigorous Approach
- Detecting communities is hard, and counting them is even harder
- Hardness Results for Signaling in Bayesian Zero-Sum and Network Routing Games