bandu@sunybcs. (Jagath SamaraBandu) (07/29/88)
I'm looking for a neural net simulator. Something on the lines of bpsim.c (BYTE Oct 87 pp155). I'd really appreciate if someone could mail me the source of bpsim.c or something similar. Many thanks in advance -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- Jagath K. Samarabandu (716)-835-4639 | bandu@cs.buffalo.edu 518, Lasalle Ave.,Buffalo,NY14215 | v092r8c2@ubvms.bitnet -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
demers@beowulf.ucsd.edu (David E Demers) (07/29/88)
In article <274@cs.Buffalo.EDU> bandu@sunybcs. (Jagath SamaraBandu) writes: >I'm looking for a neural net simulator. Something on the lines of bpsim.c (BYTE Oct 87 pp155). I'd really appreciate if someone could mail me the source of >bpsim.c or something similar. Might I be the first (possibly of many, forgive this posting if so...) to recommend volume 3 of the PDP series. I don't have it in front of me right now, but I'm positive it is entitled "Explorations in Parallel and Distributed Processing" by McClelland and Rumelhart, MIT Press, 1988. It is only $30. It contains code for back-prop as well as a number of other models. Can't beat the price! Other good simulators are SunNet (essentially only back propogation learning in feed-forward networks, but easy to use on the Sun), and the Rochester connectionist simulator (more things for the programmer to control than SunNet), available from Rochester. Sorry, don't remember the specific contact. I haven't used bpsim.c, though I do have it and am mailing it to you in a separate note. Not enough time to play with everything! > >Many thanks in advance > >-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- >Jagath K. Samarabandu (716)-835-4639 | bandu@cs.buffalo.edu >518, Lasalle Ave.,Buffalo,NY14215 | v092r8c2@ubvms.bitnet >-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- Dave DeMers demers@cs.ucsd.edu UCSD Department of CSE C-014 La Jolla, CA
brickman@cme-durer.ARPA (Jonathan E. Brickman) (07/29/88)
----- News saved at 29 Jul 88 13:14:09 GMT One very capable machine worth mentioning is Masscomp. It runs 4 or more 25 MHz 68030's, 128+ megs RAM, and X-Windows over Unix. It runs real-time Unix, which reportedly allows any or all of the CPUs to run Unix while those not running Unix dedicate themselves to computation and I/O. This machine may well outperform your hypothetical machine. It probably outprices it as well: $350K for the above system. ||Jonathan E. Brickman Newsgroups: comp.lang.idl Subject: Re: Workstation Questions Summary: Expires: References: <552@helios.ee.lbl.gov> Sender: Reply-To: brickman@rosie (Jonathan E. Brickman) Followup-To: Distribution: comp.sys.workstations,comp.edu,comp.graphics,comp.os.vms Organization: The National Bureau of Standards Keywords: workstations,masscomp One very capable machine worth mentioning is Masscomp. It runs 4 or more 25 MHz 68030's, 128+ megs RAM, and X-Windows over Unix. It runs real-time Unix, which reportedly allows any or all of the CPUs to run Unix while those not running Unix dedicate themselves to computation and I/O. This machine may well outperform your hypothetical machine. It probably outprices it as well: $350K for the above system. ||Jonathan E. Brickman Newsgroups: comp.ai.neural-nets Subject: Summary: Expires: References: <274@cs.Buffalo.EDU> <5184@sdcsvax.UCSD.EDU> Sender: brickman@rosie (Jonathan E. Brickman) Reply-To: Followup-To: Distribution: Organization: The National Bureau of Standards Keywords: neural Does anyone know if there exists a network, simulated or digital or analog hardware, which will allow a non-selforganizing network of multiple million neurons? I am looking for a very fast way to achieve image mapping capability, and so far it looks like I go with either a heavily parallel multi-CPU system or a really good neural net. ||Jonathan E. Brickman