milikich@iccgcc.decnet.ab.com (Mike Milikich, Allen-Bradley Company) (12/11/90)
Connectionists, PDPers, and whomever else is reading this : I am struggling to implement a simple Boltzmann Machine Learning algorithm. If anyone can guide me to good references which deal with learning by simulated annealing, or has some sample code which might be helpful, I would be tremendously appreciative. I have already read Chapter 7 in "Parallel Distributed Processing" Vol. I (Hinton and Sejnowski), and it only served to cloud my understanding. Email to the address below is fine - I will summarize if I get enough response. Thanks in advance!!! Mike -------------------------------------------------------------------------------- | milikich%iccgcc.decnet@consrt.rok.com |"...and the men who hold high places | | allen-bradley company | must be the ones to start..." | | 747 alpha drive |--------------------------------------| | highland heights, ohio 44143 | I don't speak for A-B, | | (216) 646-3494 FAX : (216) 646-4343 | and I know they don't speak for me. | --------------------------------------------------------------------------------
fahn@cs.columbia.edu (Paul N. Fahn) (12/11/90)
In article <2386.2763f7de@iccgcc.decnet.ab.com> milikich@iccgcc.decnet.ab.com (Mike Milikich, Allen-Bradley Company) writes: >Connectionists, PDPers, and whomever else is reading this : > > I am struggling to implement a simple Boltzmann Machine Learning >algorithm. If anyone can guide me to good references which deal with learning >by simulated annealing, or has some sample code which might be helpful, I >would be tremendously appreciative. I have already read Chapter 7 in >"Parallel Distributed Processing" Vol. I (Hinton and Sejnowski), and it only >served to cloud my understanding. Email to the address below is fine - I will >summarize if I get enough response. Thanks in advance!!! > >Mike > You can try looking at "Simulated Annealing and Boltzmann Machines", by Aarts and Korst, Wiley, 1989. But they basically give a more detailed presentation of Hinton and Sejnowski's learning algorithm, with more mathematical background. I don't know what you mean by `learning by simulated annealing'. Why don't you post (or email me) specific questions about the H&S algorithm. Perhaps I can answer them, as I've been thinking about these things lately. I've never coded it, though, so no promises. .Paul. fahn@cs.columbia.edu