Restart schedules for ensembles of problem instances

Authors:
Matthew Streeter;Daniel Golovin;Stephen F. Smith
Affiliations:
School of Computer Science, Carnegie Mellon University, Pittsburgh, PA;School of Computer Science, Carnegie Mellon University, Pittsburgh, PA;School of Computer Science, Carnegie Mellon University, Pittsburgh, PA
Venue:
AAAI'07 Proceedings of the 22nd national conference on Artificial intelligence - Volume 2
Year:
2007

Citing 9
Cited 4

Optimal speedup of Las Vegas algorithms

Information Processing Letters
Boosting combinatorial search through randomization

AAAI '98/IAAI '98 Proceedings of the fifteenth national/tenth conference on Artificial intelligence/Innovative applications of artificial intelligence
The Nonstochastic Multiarmed Bandit Problem

SIAM Journal on Computing
Restart Policies with Dependence among Runs: A Dynamic Programming Approach

CP '02 Proceedings of the 8th International Conference on Principles and Practice of Constraint Programming
Dynamic restart policies

Eighteenth national conference on Artificial intelligence
Adaptive Routing Using Expert Advice

The Computer Journal
Combining multiple heuristics online

AAAI'07 Proceedings of the 22nd national conference on Artificial intelligence - Volume 2
Learning restart strategies

IJCAI'07 Proceedings of the 20th international joint conference on Artifical intelligence
Minimizing regret with label efficient prediction

IEEE Transactions on Information Theory

Strategies for Solving SAT in Grids by Randomized Search

Proceedings of the 9th AISC international conference, the 15th Calculemas symposium, and the 7th international MKM conference on Intelligent Computer Mathematics
A self-adaptive multi-engine solver for quantified Boolean formulas

Constraints
Combining multiple heuristics online

AAAI'07 Proceedings of the 22nd national conference on Artificial intelligence - Volume 2
Learning algorithm portfolios for parallel execution

LION'12 Proceedings of the 6th international conference on Learning and Intelligent Optimization

Quantified Score

Hi-index	0.00

Visualization

Abstract

The mean running time of a Las Vegas algorithm can often be dramatically reduced by periodically restarting it with a fresh random seed. The optimal restart schedule depends on the Las Vegas algorithm's run length distribution, which in general is not known in advance and may differ across problem instances. We consider the problem of selecting a single restart schedule to use in solving each instance in a set of instances. We present offline algorithms for computing an (approximately) optimal restart schedule given knowledge of each instance's run length distribution, generalization bounds for learning a restart schedule from training data, and online algorithms for selecting a restart schedule adaptively as new problem instances are encountered.