Results 1 
6 of
6
Learning Evaluation Functions to Improve Optimization by Local Search
 Journal of Machine Learning Research
, 2000
"... This paper describes algorithms that learn to improve search performance on largescale optimization tasks. The main algorithm, Stage, works by learning an evaluation function that predicts the outcome of a local search algorithm, such as hillclimbing or Walksat, from features of states visited durin ..."
Abstract

Cited by 57 (0 self)
 Add to MetaCart
This paper describes algorithms that learn to improve search performance on largescale optimization tasks. The main algorithm, Stage, works by learning an evaluation function that predicts the outcome of a local search algorithm, such as hillclimbing or Walksat, from features of states visited during search. The learned evaluation function is then used to bias future search trajectories toward better optima on the same problem. Another algorithm, XStage, transfers previously learned evaluation functions to new, similar optimization problems. Empirical results are provided on seven largescale optimization domains: binpacking, channel routing, Bayesian network structurefinding, radiotherapy treatment planning, cartogram design, Boolean satisfiability, and Boggle board setup.
A Fast Heuristic Algorithm Based on Verification and Elimination Methods for Maximum Clique Problem
"... A clique in an undirected graph G = (V, E) is a subset V ' ⊆ V of vertices, each pair of which is connected by an edge in E. The clique problem is an optimization problem of finding a clique of maximum size in graph. The clique problem is NPComplete. We have succeeded in developing a fast alg ..."
Abstract
 Add to MetaCart
A clique in an undirected graph G = (V, E) is a subset V ' ⊆ V of vertices, each pair of which is connected by an edge in E. The clique problem is an optimization problem of finding a clique of maximum size in graph. The clique problem is NPComplete. We have succeeded in developing a fast algorithm for maximum clique problem by employing the method of verification and elimination. For a graph of size N there are 2 N sub graphs, which may be cliques and hence verifying all of them, will take a long time. Idea is to eliminate a major number of sub graphs, which cannot be cliques and verifying only the remaining sub graphs. This heuristic algorithm runs in polynomial time and executes successfully for several examples when applied to random graphs and DIMACS benchmark graphs. 1.
The MIT Press
"... Published one article at a time in L ATEX source form on the Internet. Pagination ..."
Abstract
 Add to MetaCart
Published one article at a time in L ATEX source form on the Internet. Pagination
Oscillations in Discrete and Continuous Hopfield Networks
"... This chapter is neatly partitioned into two parts: one dealing with oscillations in discrete Hopfield networks and the other with oscillations in continuous Hopfield networks. The single theme spanning both parts is that of the Hopfield model and its energy function. The first part deals with analyz ..."
Abstract
 Add to MetaCart
This chapter is neatly partitioned into two parts: one dealing with oscillations in discrete Hopfield networks and the other with oscillations in continuous Hopfield networks. The single theme spanning both parts is that of the Hopfield model and its energy function. The first part deals with analyzing oscillations in the discrete Hopfield network with symmetric weights, and speculating on possible uses of such behavior. By imposing certain restrictions on the weights, an exact characterization of the oscillatory behavior is obtained. Possible uses of this characterization are examined. The second part deals with injecting chaotic or periodic oscillations into continuous Hopfield networks, for the purposes of solving optimization problems. When the continuous Hopfield model is used to solve an optimization problem, the results are often mediocre because of the convergent nature of its dynamical algorithm. To circumvent this limitation, we develop mechanisms for injecting controllable c...
Performance of Neural Network Algorithms for Maximum Clique On Highly Compressible Graphs
"... 0 Maximum Clique perform on graphs drawn from q(x), as compared to those drawn from u(n). The experimental results are as follows. All nine algorithms we evaluated performed roughly equallywell on u(n), where as three of themthe simplest onesperformed markedly poorer than the other six on q(x). ..."
Abstract
 Add to MetaCart
0 Maximum Clique perform on graphs drawn from q(x), as compared to those drawn from u(n). The experimental results are as follows. All nine algorithms we evaluated performed roughly equallywell on u(n), where as three of themthe simplest onesperformed markedly poorer than the other six on q(x). Our results suggest that q(x), while postulated as a more realistic distribution to test the performance of algorithms than u(n), might also discriminate their performance better. Our q(x) sampler can be used to generate compressible instances of any discrete problem. 1
Learning Evaluation Functions to Improve Local Search
"... This paper describes Stage, a learning algorithm that automatically improves search performance on largescale optimization problems. Stage learns an evaluation function that predicts the outcome of a local search algorithm, such as hillclimbing or Walksat, from features of states visited during sea ..."
Abstract
 Add to MetaCart
This paper describes Stage, a learning algorithm that automatically improves search performance on largescale optimization problems. Stage learns an evaluation function that predicts the outcome of a local search algorithm, such as hillclimbing or Walksat, from features of states visited during search. The learned evaluation function is used to bias future search trajectories toward better optima on the same problem. This paper presents the Stage algorithm; an extension, XStage, that transfers learned evaluation functions to new, similar optimization problems; and empirical results on seven largescale optimization domains: binpacking, channel routing, Bayes network structurefinding, radiotherapy treatment planning, cartogram design, Boolean satisfiability, and Boggle board setup.