A Q'tron Neural-Network Approach to Solve the Graph Coloring Problems

This paper proposes a novel methodology to solve the graph coloring problem (GCP) using the Q'tron neural- network (NN) model. The Q'tron NN for GCP will be built as a known-energy system. This can make the NN local- minima-free and perform the so-called goal-directed search. Consider k-GCP as a goal to solve a GCP using at most k different colors. By continuously refining our goal, i.e., decreasing the value k, we can then 'demand' the NN to fulfill better and better goals progressively. Experiments using DI-MACS benchmarks were done using such an approach, and comparison was made with the DSATUR algorithm. The result supports the soundness of our approach.

[1]  Gerd Stumme,et al.  Mining frequent patterns with counting inference , 2000, SKDD.

[2]  E. Salari,et al.  An ACO algorithm for graph coloring problem , 2005, 2005 ICSC Congress on Computational Intelligence Methods and Applications.

[3]  Habiba Drias,et al.  Ant Colony System for Graph Coloring Problem , 2005, International Conference on Computational Intelligence for Modelling, Control and Automation and International Conference on Intelligent Agents, Web Technologies and Internet Commerce (CIMCA-IAWTIC'06).

[4]  Li-Chen Fu,et al.  A complete neural network approach to solving a class of combinatorial problems , 1991, IJCNN-91-Seattle International Joint Conference on Neural Networks.

[5]  Mohammed J. Zaki,et al.  Efficiently mining maximal frequent itemsets , 2001, Proceedings 2001 IEEE International Conference on Data Mining.

[6]  Heikki Mannila,et al.  Levelwise Search and Borders of Theories in Knowledge Discovery , 1997, Data Mining and Knowledge Discovery.

[7]  Hongjun Lu,et al.  Efficient Search of Reliable Exceptions , 1999, PAKDD.

[8]  Keun Ho Ryu,et al.  Mining association rules on significant rare data using relative support , 2003, J. Syst. Softw..

[9]  G Siest,et al.  Objectives, Design and Recruitment of a Familial and Longitudinal Cohort for Studying Gene-Environment Interactions in the Field of Cardiovascular Risk: The Stanislas Cohort , 1998, Clinical chemistry and laboratory medicine.

[10]  Alain Hertz,et al.  Using tabu search techniques for graph coloring , 1987, Computing.

[11]  Chirag M. Patel,et al.  An ant-based algorithm for coloring graphs , 2008, Discret. Appl. Math..

[12]  Marzena Kryszkiewicz.  Concise representation of frequent patterns based on disjunction-free generators , 2001, Proceedings 2001 IEEE International Conference on Data Mining.

[13]  C. D. Gelatt,et al.  Optimization by Simulated Annealing , 1983, Science.

[14]  Vladimir Gurvich,et al.  On Maximal Frequent and Minimal Infrequent Sets in Binary Matrices , 2003, Annals of Mathematics and Artificial Intelligence.

[15]  Cecilia R. Aragon,et al.  Optimization by Simulated Annealing: An Experimental Evaluation; Part II, Graph Coloring and Number Partitioning , 1991, Oper. Res..

[16]  Einoshin Suzuki,et al.  Undirected Discovery of Interesting Exception Rules , 2002, Int. J. Pattern Recognit. Artif. Intell..

[17]  Wynne Hsu,et al.  Mining association rules with multiple minimum supports , 1999, KDD '99.

[18]  Gary M. Weiss.  Mining with rarity: a unifying framework , 2004, SKDD.

[19]  Howard A. Peelle.  Graph coloring in J: an introduction , 2000, ACM SIGAPL APL Quote Quad.

[20]  Ramesh C Agarwal,et al.  Depth first generation of long patterns , 2000, KDD '00.