Secure Approximation Guarantee for Cryptographically Private Empirical Risk Minimization.
Toshiyuki TakadaHiroyuki HanadaYoshiji YamadaJun SakumaIchiro TakeuchiPublished in: ACML (2016)
Keyphrases
- empirical risk minimization
- approximation guarantees
- greedy algorithm
- statistical learning theory
- vc dimension
- approximation algorithms
- uniform convergence
- approximation ratio
- empirical risk
- np hard
- computationally tractable
- lower bound
- linear programming relaxation
- rates of convergence
- phase transition
- generalization bounds
- objective function
- machine learning
- upper bound
- theoretical framework
- np complete
- multi class