Teaching and compressing for low VC-dimension.
Shay MoranAmir ShpilkaAvi WigdersonAmir YehudayoffPublished in: CoRR (2015)
Keyphrases
- vc dimension
- upper bound
- sample size
- concept classes
- covering numbers
- lower bound
- sample complexity
- vapnik chervonenkis dimension
- empirical risk minimization
- statistical learning theory
- distribution free
- inductive inference
- generalization bounds
- worst case
- concept class
- compression scheme
- pac learnability
- pac learning
- data compression
- learning environment
- uniform convergence
- e learning
- euclidean space
- computational complexity
- learning process
- feature space
- data sets