Almost all monotone Boolean functions are polynomially learnable using membership queries.
Ilya ShmulevichAleksey D. KorshunovJaakko AstolaPublished in: Inf. Process. Lett. (2001)
Keyphrases
- monotone boolean functions
- membership queries
- dnf formulas
- concept class
- uniform distribution
- pac learning
- exact learning
- boolean functions
- target concept
- equivalence queries
- efficient learning
- concept classes
- learning algorithm
- query complexity
- agnostic learning
- dnf formulae
- polynomial size
- target function
- monotone dnf
- membership and equivalence queries
- equivalence and membership queries
- read once formulas
- computational learning theory
- statistical queries
- monotone dnf formulas
- term dnf
- remains np hard
- pattern languages
- positive and negative