Improved Inapproximability of VC Dimension and Littlestone's Dimension via (Unbalanced) Biclique.
Pasin ManurangsiPublished in: CoRR (2022)
Keyphrases
- vc dimension
- concept classes
- vapnik chervonenkis
- sample complexity
- vapnik chervonenkis dimension
- upper bound
- concept class
- sample size
- mistake bound
- lower bound
- decision lists
- statistical learning theory
- pac learning
- worst case
- generalization bounds
- covering numbers
- inductive inference
- learning theory
- theoretical analysis
- learning algorithm
- supervised learning
- learning models
- pac model
- efficient learning
- euclidean space
- compression scheme
- learning problems
- generalization error
- noise tolerant
- np hard
- uniform distribution