!!Extended neighborhoods (1121209), страница 4
Текст из файла (страница 4)
This general model uniesvarious SA's within the same framework and allows derivation of many useful resultswith both theoretical and practical importance.A practical method of using continuous probability density functions as SA's generation functions to extend SA's neighbourhood is given herein. Such SA's reducethe computation time and outperform SSA. This is also shown by our experimental results [24], where we used TSP as an example and compared the result of SSAand that of SA with the Normal function as generation function.
A question whicharises in this stage is which probabiity density function can give SA best performance.We have done some preliminary experiments on this issue [24]. We compared SA'swith ve dierent generation functions, i.e., Normal function, exponential function,Cauchy function, stable function with index 21 and a hybrid function with Cauchy athigh temperatures and Normal at low temperatures, where the density function ofthe stable distribution with index 21 is [25](66)f (x) = p 1 3 exp ? 21x ; x > 02xIt is easy to show that the expectation and variance of (66) do not exist. Our resultsindicate that the hybrid function and Cauchy function are better than the other threeand the stable function with index 21 is the worst.
However, more extensive empiricalstudy and rigorous theoretical analysis are needed to nd out which kinds of distribution or combinations of distributions are best for SA's generation and acceptancefunctions.Another important research direction is to incorporate the concept of knowledgeguided search into SA or other stochastic search methods [26].
It is very dicultfor SA to know whether a region in the conguration space has been explored orwhether a region is good, even though SA has been wandering in the space for a longtime. SA should be able to discover some hints or gain some knowledge about thespace through previous searches to guide the subsequent searches. The totally blindstochastic search can hardly achieve very high eciency from the AI's point of view.Acknowledgement | The author would like to thank Profs.
Richard P. Brentand Guo-Jie Li for clarifying the meaning of Theorem 3.1 and Dr. Iain D. G. Macleodfor correcting errors in English .Xin Yao: Simulated Annealing with Extended Neighbourhood20References[1] D. F. Wong, H. W. Leong, and C. L. Liu.Simulated Annealing for VLSI Design.Kluwer Academic Publishers, Boston, 1988.[2] S. Kirkpatrick, C. D. Gelatt, and M. P.
Vecchi.Optimization by simulated annealing.Science, 220:671{680, 1983.[3] P. J. M. van Laarhoven and E. H. L. Aarts.Simulated Annealing: Theory and Applications.D. Reidel Publishing Co., 1987.[4] E. H. L. Aarts and P. J. M. van Laarhoven.Statistical cooling: a general approach to combinatorial optimization problems.Philips Journal of Research, 40:193{226, 1985.[5] D. H. Ackley, G.
E. Hinton, and T. J. Sejnowski.A learning algorithm for Boltzmann machine.Cognitive Science, 9:147{169, 1985.[6] M. F. Tenorio and W. T. Lee.Self-organizing neural nets for the identication problem.In D. S. Touretzky, editor, Advances in Neural Information Processing Systems(I). Morgan Kaufmann, San Mateo, CA, 1989.[7] P.
Carnevalli, L. Coletti, and S. Petarnello.Image processing by simulated annealing.IBM Journal of Research and Development, 29:569{579, 1985.[8] E. D. Sontag and H. J. Sussman.Image restoration and segmentation using the annealing algorithm.In Proceedings of the 24th International Conference on Decision and Control,pages 768{773, Ft. Lauderdale, Florida, 1985.[9] A. A. El Gemal.Using simulated annealing to design good codes.IEEE Trans. on Information Theory, IT-33:116{123, 1987.[10] D.
H. Ackley.An empirical study of bit vector function optimization.In L. Davis, editor, Genetic Algorithms and Simulated Annealing, chapter 13,pages 170{204. Morgan Kaufmann, San Mateo, CA, 1987.Xin Yao: Simulated Annealing with Extended Neighbourhood21[11] N. Metropolis, A. Rosenbluth, M.
Rosenbluth, A. Teller, and E. Teller.Equations of state calculations by fast computing machines.Journal of Chemical Physics, 21:1087{1091, 1953.[12] V. C erny.Thermodynamical approach to the travelling salesman problem: an ecient simulation algorithm.Journal of Optimization Theory and Applications, 45:41{51, 1985.[13] D. L.
Isaacson and R. W. Madsen.Markov Chains Theory and Applications.John Wiley & Sons, 1976.[14] S. Geman and D. Geman.Stochastic relaxation, Gibbs distribution, and the Bayesian restoration of images.IEEE Trans. on Pattern Analysis and Machine Intelligence, PAMI-6:721{741,1984.[15] B. Hajek.Cooling schedules for optimal annealing.Mathematics of Operations Research, 13:311{329, 1988.[16] D. Mitra, F. Romeo, and A. Sangiovanni-Vincentelli.Convergence and nite-time behavior of simulated annealing.Advanced Applied Probability, 18:747{771, 1986.[17] X. Yao and G.-J. Li.Performance analysis of simulated annealing.In L.-W.
Chen, B.-C. Chen, and Q.-N. Sun, editors, Proc. of the Third PanPacic Computer Conf., pages 972{980, Beijing, 1989. International Academic Publishers and Pergamon Press.[18] M. P. Vecchi and S. Kirkpatrick.Global wiring by simulated annealing.IEEE Trans. on CAD, CAD-2:215{222, 1983.[19] S. Anily and A. Federgruen.Simulated annealing methods with general acceptance probabilities.Journal of Applied Probability, 24:657{667, 1987.[20] I. O. Bohachevsky, M. E.
Johnson, and M. L. Stein.Generalized simulated annealing for function optimization.Technometrics, 28:209{217, 1986.[21] H. L. Royden.Xin Yao: Simulated Annealing with Extended Neighbourhood[22][23][24][25][26]22Real Analysis.Macmillan Publishing Co., Inc., 2nd edition, 1968.S. Anily and A. Federgruen.Ergodicity in parametric nonstationary Markov chains: an application to annealing methods.Operations Research, 35:867{874, 1987.B. Gidas.Non-stationary Markov chains and convergence of the annealing algorithm.Journal of Statistical Physics, 39:73{131, 1985.X.
Yao.A General Stochastic Search Algorithm | Theory and Applications of SimulatedAnnealing.PhD thesis, University of Science and Technology of China, Hefei, Anhui 230026,P.R.China, 1989.W. Feller.An Introduction to Probability Theory and Its Applications, volume 2.John Wiley & Sons, Inc., 2nd edition, 1971.X. Yao.Optimization by genetic annealing.In M. Jabri, editor, Proc. of Second Australian Conf. on Neural Networks, pages94{97, Sydney, Australia, 1991..
![](https://s.studizba.com/z.php?f=/templates/si/i/noavatar.png&w=100&h=100&t=1"){kind=avatar}
