Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!ames!apple!brutus.cs.uiuc.edu!wuarchive!kuhub.cc.ukans.edu!rezac From: REZAC@kuhub.cc.ukans.edu Newsgroups: comp.ai.neural-nets Subject: Re: Neural Nets for PCs? (summary) Message-ID: <8860@kuhub.cc.ukans.edu> Date: 17 Aug 89 15:33:34 GMT References: <8024@kuhub.cc.ukans.edu> <6935@sdcsvax.UCSD.Edu> Organization: University of Kansas Academic Computing Services Lines: 109 The following are responses to my request for implementations of simple PC based neural networks. Thanks to everyone who responded and I hope this information is as beneficial for everyone who reads this as it was for me. Return-path: plonski@aerospace.aero.org *I would suggest the following handbook which comes with C-source code that run on MS-C for the *PC. It is the workbook that goes with the PDP Vol. 1 book which is the standard in the field. *The workbook goes through the code and is an excellent intro to NN. *Good Luck. I have also included the reference for the PDP Vol 1. %A J.L. McClelland %A D.E. Rumelhart %T Explorations in Parallel Distributed Processing %o A Handbook of Models, Programs, and Exercises %I |MIT| %D 1988 %K PDP, computer programs %X Interactive Activation \& Competition 11 %A D. E. Rumelhart %A J. L. McClelland %A and the PDP\ Research\ Group %T Parallel Distributed Processing %o Vol. 1: Foundations %I |MIT| %D 1986 %K PDP-V1, NN, PDP, Parallel Processing, Widrow Hoff, Hebb %X Key Reference book for Parallel Processing and NN Another person responded . . . . *How simple are you looking for and what are you planning *to use them for????? I have several programs, which were *originally done in fortran, and I recoded into c that I got *in a neural nets course I took last spring. They include *a Hopfield net, an unsupervised net, a preceptron and a one *hidden layer feedforward net. If you are interested let me *know and I will check with my Prof. to make sure its ok to *give them out. Rather than have his mailer be swamped by requests for the code, I'll wait til I hear whether Ken can distribute it or not. If he can I'll see if it can be made available via ftp. Return-path: ncegeber@plains.NoDak.edu *I don't know if these will all fit your criteria, but I have found these *sources: *Dr. Dobb's Journal, Jan. 1989 had 2 articles about NN's. Source code was *in C. I got a copy from a local BBS, but they (Dr. Dobb's Journal) are *supposed to have code available on Compuserve, and also on a dial-up service *at 603-882-1599. I haven't used their dial-up service, but this is what it *says in the mag. Supports 300/1200/2400 baud, 8 data bits, no parity, 1 *stop bit. When it answers, respond with 'listings' to the login prompt. *It is supposed to have listings back through the beginning of the year. *This code compiles/runs with Turbo C. These comprise a fair introduction, source can also be found in pd1: on simtel20.arpa *Another source I found is AI Expert magazine's BBS. I have downloaded the *source from the June 1988 issue. It is code for a backprop NN in C. It *was written to run on the MacIntosh, but I had little trouble converting it *to Turbo C. There are a number of BBS's that have code listings from AI *Expert mag. See any issue of AI Expert (or Computer Language - the BBS's *have code for both mags.). They (AI Expert) have had a series of articles *called 'Neural Networks Primer' so they may have code for other NN's. Code for this magazine can also be found on simtel in pd1: *Byte had an article on backprop in the Oct 1987 issue. The last time I logged *on to Byte's listing service, they didn't have listings going back that far *so I don't know if it is available. Maybe it is still available on BIX, for *those that subscribe to it. *A recent issue (June/July 1989, I think) of Circuit Cellar, INK had some *articles about NN's. I downloaded FORTRAN code for backprop from their BBS. *It is at (203) 871-1988, 8 data bits, no parity, 1 stop bit. I haven't *tried to compile or run it yet. *One more thing I just remembered, some time in the last several months, *some Xlisp code was posted in comp.sources.amiga. I tried it with Xlisp on *the PC and it seemed to run okay. I have thought about trying to convert it *to C++, but haven't found the time yet. From: heirich@beowulf.ucsd.edu (Alan Heirich) *Try volume three of "Parallel Distributed Processing" by Rumelhart & *McClelland (MIT Press). It consists of C source code and sample programs *for various neural net learning algorithms and is intended for educational *use. It comes with a disk for IBM PC compatibles. One final reference which I ran across in the bookstore last night - Computer Simulation in Brain Science - Cambridge University Press - $65.00 I looked through it and it appears to cover most of the topics I'm interested in, though I haven't purchased the book yet. If anyone has read the book and wants to comment, please do. Disclaimer: It's mine, but if you want it you can have it. *=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=* CHARLES REZAC * bitnet: REZAC@UKANVAX internet: rezac@kuhub.cc.ukans.edu