A Sharp Lower Bound for Agnostic Learning with Sample Compression Schemes.
Steve HannekeAryeh KontorovichPublished in: ALT (2019)
Keyphrases
- compression scheme
- agnostic learning
- lower bound
- concept class
- vc dimension
- upper bound
- image compression
- pac learning
- uniform distribution
- concept classes
- data compression
- sample size
- learning theory
- compression ratio
- compression algorithm
- membership queries
- noise tolerant
- lower and upper bounds
- sample complexity
- objective function
- worst case
- target function
- optimal solution
- dnf formulas
- upper and lower bounds
- statistical queries
- high quality
- concept learning
- noisy data
- information theoretic
- model selection
- text classification
- binary classification problems
- multi class
- computational learning theory