Not All Learnable Distribution Classes are Privately Learnable.
Mark BunGautam KamathArgyris MouzakisVikrant SinghalPublished in: CoRR (2024)
Keyphrases
- recursive functions
- positive data
- learning from positive data
- membership and equivalence queries
- hypothesis space
- pac learnability
- dnf formulas
- membership queries
- pac learning
- learning algorithm
- concept class
- exact learning
- pattern languages
- neural network
- decision lists
- power law
- efficient learning
- random variables
- multi class
- class membership
- positive examples
- spatial distribution
- probability distribution
- function classes
- training data
- probability density function
- sample size