Polynomial learnability of stochastic rules with respect to the KL-divergence and quadratic distance

IEICE Transactions on Information and Systems

Volumn E84-D, Issue 3, 2001, Pages 290-316

  Abe, Naoki ^a   Takeuchi, Jun Ichi ^a   Warmuth, K Manfred ^a  

^a NEC CORPORATION   (Japan)

Author keywords

KL divergence; P concepts; PAC learning; Quadratic distance; Stochastic rules

Indexed keywords

COMPUTATIONAL COMPLEXITY; COMPUTER SIMULATION; CONFORMAL MAPPING; FINITE AUTOMATA; MATHEMATICAL MODELS; MAXIMUM LIKELIHOOD ESTIMATION; OPTIMIZATION; POLYNOMIALS; PROBABILITY DISTRIBUTIONS; RANDOM PROCESSES; SET THEORY;

KULLBACK-LEIBLER DIVERGENCE; POLYNOMIAL TIME LEARNABILITY; QUADRATIC DISTANCE; STOCHASTIC RULES;

LEARNING ALGORITHMS;

EID: 0035281455     PISSN: 09168532     EISSN: None     Source Type: Journal    
DOI: None     Document Type: Article

References (18)

1. Learning from noisy examples
2. Polynomial learnability of probabilistic concepts with respect to the Kullback-Leibler divergence
3. On the computational complexity of approximating probability distributions by probabilistic automata
4. Decision theoretic generalizations of the PAC model for neural net and other learning applications
5. A comparison of new and old algorithms for a mixture estimation problem
6. Efficient distribution-free learning of probabilistic concepts
7. Toward efficient agnostic learning
8. A lower bound for discrimination in terms of variation
9. On-line learning of linear functions and iterative solution of linear systems
10. The gradient projection method for nonlinear programming. Part I. Linear constraints
11. The design and analysis of efficient learning algorithms
12. A theory of the learnable
13. Inductive principles of the search for empirical dependences (methods based on weak convergence of probability measures)
14. A learning criterion for stochastic rules