Agnostic Learnability of Halfspaces via Logistic Loss.
Ziwei JiKwangjun AhnPranjal AwasthiSatyen KaleStefani KarpPublished in: ICML (2022)
Keyphrases
- dnf formulas
- agnostic learning
- concept class
- membership queries
- decision lists
- exact learning
- uniform distribution
- concept classes
- term dnf
- vapnik chervonenkis dimension
- boolean functions
- equivalence queries
- vc dimension
- pac learning
- upper bound
- upper and lower bounds
- learning theory
- finite automata
- target concept
- pac learnability
- target function
- monotone dnf formulas
- lower bound
- sample complexity
- uniform convergence
- pattern languages
- sufficient conditions
- half spaces
- real valued functions
- inductive inference
- noise tolerant
- boosting algorithms
- conjunctive queries
- query complexity
- concept learning
- tree patterns