Using Prediction to Improve Combinatorial Optimization Search

Justin A. Boyan, Andrew W. Moore
Proceedings of the Sixth International Workshop on Artificial Intelligence and Statistics, PMLR R1:55-66, 1997.

Abstract

This paper describes a statistical approach to improving the performance of stochastic search algorithms for optimization. Given a search algorithm $A$, we learn to predict the outcome of $A$ as a function of state features along a search trajectory. Predictions are made by a function approximator such as global or locally-weighted polynomial regression; training data is collected by Monte-Carlo simulation. Extrapolating from this data produces a new evaluation function which can bias future search trajectories toward better optima. Our implementation of this idea, STAGE, has produced very promising results on two large-scale domains.

Cite this Paper


BibTeX
@InProceedings{pmlr-vR1-boyan97a, title = {Using Prediction to Improve Combinatorial Optimization Search}, author = {Boyan, Justin A. and Moore, Andrew W.}, booktitle = {Proceedings of the Sixth International Workshop on Artificial Intelligence and Statistics}, pages = {55--66}, year = {1997}, editor = {Madigan, David and Smyth, Padhraic}, volume = {R1}, series = {Proceedings of Machine Learning Research}, month = {04--07 Jan}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/r1/boyan97a/boyan97a.pdf}, url = {https://proceedings.mlr.press/r1/boyan97a.html}, abstract = {This paper describes a statistical approach to improving the performance of stochastic search algorithms for optimization. Given a search algorithm $A$, we learn to predict the outcome of $A$ as a function of state features along a search trajectory. Predictions are made by a function approximator such as global or locally-weighted polynomial regression; training data is collected by Monte-Carlo simulation. Extrapolating from this data produces a new evaluation function which can bias future search trajectories toward better optima. Our implementation of this idea, STAGE, has produced very promising results on two large-scale domains.}, note = {Reissued by PMLR on 30 March 2021.} }
Endnote
%0 Conference Paper %T Using Prediction to Improve Combinatorial Optimization Search %A Justin A. Boyan %A Andrew W. Moore %B Proceedings of the Sixth International Workshop on Artificial Intelligence and Statistics %C Proceedings of Machine Learning Research %D 1997 %E David Madigan %E Padhraic Smyth %F pmlr-vR1-boyan97a %I PMLR %P 55--66 %U https://proceedings.mlr.press/r1/boyan97a.html %V R1 %X This paper describes a statistical approach to improving the performance of stochastic search algorithms for optimization. Given a search algorithm $A$, we learn to predict the outcome of $A$ as a function of state features along a search trajectory. Predictions are made by a function approximator such as global or locally-weighted polynomial regression; training data is collected by Monte-Carlo simulation. Extrapolating from this data produces a new evaluation function which can bias future search trajectories toward better optima. Our implementation of this idea, STAGE, has produced very promising results on two large-scale domains. %Z Reissued by PMLR on 30 March 2021.
APA
Boyan, J.A. & Moore, A.W.. (1997). Using Prediction to Improve Combinatorial Optimization Search. Proceedings of the Sixth International Workshop on Artificial Intelligence and Statistics, in Proceedings of Machine Learning Research R1:55-66 Available from https://proceedings.mlr.press/r1/boyan97a.html. Reissued by PMLR on 30 March 2021.

Related Material