A Reinforcement Learning Approach to Interval Constraint Propagation

Authors:
Frédéric Goualard;Christophe Jermann
Affiliations:
LINA, CNRS FRE 2729, University of Nantes, Nantes cedex 3, France BP 92208 F-44322;LINA, CNRS FRE 2729, University of Nantes, Nantes cedex 3, France BP 92208 F-44322
Venue:
Constraints
Year:
2008

Citing 12
Cited 0

CLP(intervals) revisited

ILPS '94 Proceedings of the 1994 International Symposium on Logic programming
Iterative solution of nonlinear equations in several variables

Iterative solution of nonlinear equations in several variables
A conservative scheme for parallel interval narrowing

Information Processing Letters
Numerical Solution of Nonlinear Equations

ACM Transactions on Mathematical Software (TOMS)
On Algorithms for Obtaining a Maximum Transversal

ACM Transactions on Mathematical Software (TOMS)
Interval arithmetic: From principles to implementation

Journal of the ACM (JACM)
Reinforcement Learning

Reinforcement Learning
The Nonstochastic Multiarmed Bandit Problem

SIAM Journal on Computing
Accelerating filtering techniques for numeric CSPs

Artificial Intelligence
Gambling in a rigged casino: The adversarial multi-armed bandit problem

FOCS '95 Proceedings of the 36th Annual Symposium on Foundations of Computer Science
On considering an interval constraint solving algorithm as a free-steering nonlinear Gauss-Seidel procedure

Proceedings of the 2005 ACM symposium on Applied computing
On the selection of a transversal to solve nonlinear systems with interval arithmetic

ICCS'06 Proceedings of the 6th international conference on Computational Science - Volume Part I

Quantified Score

Hi-index	0.00

Visualization

Abstract

When solving systems of nonlinear equations with interval constraint methods, it has often been observed that many calls to contracting operators do not participate actively to the reduction of the search space. Attempts to statically select a subset of efficient contracting operators fail to offer reliable performance speed-ups. By embedding the recency-weighted average Reinforcement Learning method into a constraint propagation algorithm to dynamically learn the best operators, we show that it is possible to obtain robust algorithms with reliable performances on a range of sparse problems. Using a simple heuristic to compute initial weights, we also achieve significant performance speed-ups for dense problems.