GX.RAR@forsythe.stanford.edu (Buc Richards) (05/13/89)
Help, I am looking for Boltzmann machines to anneal! Why you ask? I am interested in the simulated annealing properties of Boltzmann machines, that is I am attempting to determine ways to create an annealing schedule by more than just trial and error or some of the suggested slow schedules reviewed in Van Laarhoven (Simulated Annealing: Theory and Applications, 1987). However, most of the literature I have read on Boltzmann machines are concerned with the learning phase or give examples that are too simple. What I would like is a Boltzmann machine who's weights have already been learned that encodes a difficult problem. Something in the range greater than 50 to 100 nodes. Any suggestions on where I could find such a beast? If you have one that I could anneal I would be happy to share my results, actually I would be happy to share my results in any case. Thanks. Cray Y-MP**ETA 10-E**Cray X-MP**NEC SX-2**Hitachi S-820/80**Cray 2S**NAS AS/XL ------------------------------------------------------------------------------- Rob Richards Internet: buc@jessica.Stanford.Edu Mechanical Engineering Dept. Bitnet: buc%jessica.Stanford.Edu@Stanford Supercomputing Support Staff Mialnet: buc%jessica.Stanford.Edu@Mit-Multics Stanford University FAX: (415) 725-8240 ------------------------------------------------------------------------------- Fujitsu VP-400**Amdahl 1400**Cyber 205**Ardent Titan**IBM 3090**Siemens H120F