neuron-request@HPLMS2.HPL.HP.COM ("Neuron-Digest Moderator Peter Marvit") (06/14/90)
Neuron Digest Wednesday, 13 Jun 1990 Volume 6 : Issue 40 Today's Topics: REQ: teaching Hopfield/graded response network Hopfield Networks -- How Best to Run? Re: Hopfield Networks -- How Best to Run? Re: Hopfield Networks -- How Best to Run? neural nets that "count" applications in chemistry About Hopfield neural-net Re: About Hopfield neural-net Library Circulation of Neural Network Journals A GA Tutorial and a GA Short Course Quantum devices for Neural Nets? Special Issue Announcement Re: Neuron Digest V6 #21 Another Request for NN Software Request for biological NN info. Cognitive Science Society Meeting Send submissions, questions, address maintenance and requests for old issues to "neuron-request@hplabs.hp.com" or "{any backbone,uunet}!hplabs!neuron-request" Use "ftp" to get old issues from hplpm.hpl.hp.com (15.255.176.205). ------------------------------------------------------------ Subject: REQ: teaching Hopfield/graded response network From: smagt@cs.vu.nl (Smagt v der PPP) Organization: V.U. Informatica, Amsterdam, the Netherlands Date: 12 Dec 89 12:14:22 +0000 I want to use the Hopfield network [1] with graded response for classification purposes, but I have trouble finding an appropriate learning rule. Are there any articles on this? Does anyone have experience? Patrick van der Smagt Reference: [1] %A J. J. Hopfield %D May 1984 %J Proceedings of the National Academy of Sciences %P 3088-3092 %T Neurons with graded response have collective computational %T properties like those of two-state neurons %V 81 ------------------------------ Subject: Hopfield Networks -- How Best to Run? From: loren@tristan.llnl.gov (Loren Petrich) Organization: Lawrence Livermore National Laboratory Date: 23 Mar 90 05:00:33 +0000 I'm rather new to Neural Networks, so I'd like some advice. I'm constructing a simulation of a Hopfield NN, which consists of minimizing the following energy function: E = 1/2*sum(i,j)T(i,j)S(i)S(j) where the nodes have S(i) equal to +/- 1 and the T's are the weights. I'm planning to use simulated annealing for doing this, and I wonder what is the most suitable schedule -- what behavior of temperature over time. If anyone knows of a superior algorithm, I would love to learn about it. I wonder if this is an NP-complete problem. There is a straightforward algorithm for doing this, but its time is proportional to 2^n for dimension n. I think that simulated annealing may be able to do a lot better; perhaps n^2. ^ Loren Petrich, the Master Blaster \ ^ / loren@sunlight.llnl.gov \ ^ / One may need to route through any of: \^/ <<<<<<<<+>>>>>>>> lll-lcc.llnl.gov /v\ lll-crg.llnl.gov / v \ star.stanford.edu / v \ v "What do you MEAN it's not in the computer?!?" -- Madonna ------------------------------ Subject: Re: Hopfield Networks -- How Best to Run? From: russell@minster.york.ac.uk Organization: Department of Computer Science, University of York, England Date: 27 Mar 90 18:36:52 +0000 > I'm rather new to Neural Networks, so I'd like some advice. aren't we all :) >I'm constructing a simulation of a Hopfield NN, which consists of (with simulated annealing, it's really the Boltzmann machine) > I'm planning to use simulated annealing for doing this, and I >wonder what is the most suitable schedule -- what behavior of >temperature over time. Hmmm, this is tricky. The behaviour of the system is difficult to characterise accurately, and so the production of an ``optimal'' annealing schedule is difficult. As you probably know, the point about simulated annealing is to allow local minima to be escaped from by allowing jumps to higher energy levels. Now, there comes a stage in the annealing schedule where jumps to lower enegry states occur much more frequently than jumps to high energy states, and this ``transition'' period, which bears resemblances to state transitions in solids/liquids, should be held for as long as possible. It is easiest to see on a graph so here goes... |+ | + | + | + | + | + | + | + | + | + | + | + | + _________________________ a b where the vertical axis is the average energy of the system, <E>, and the horizontal axis is the temperature. The best nets will have spend most time with the temperature in the range a-b on the graph, where the energy is decreasing quickest. Now, I suppose that you could do a gradient analysis of the global energy function to determine your annealing schedule, but that `feels' messy, since it involves a global summation and removes the principle of local computation that the Boltzmann net embodies. Has anyone *tried* a global calculation? Russell. ____________________________________________________________ Russell Beale, Advanced Computer Architecture Group, Dept. of Computer Science, University of York, Heslington, YORK. YO1 5DD. UK. Tel: [044] (0904) 432762 russell@uk.ac.york.minster JANET russell%minster.york.ac.uk@nsfnet-relay.ac.uk ARPA ..!mcsun!ukc!minster!russell UUCP russell@minster.york.ac.uk eab mail ____________________________________________________________ ------------------------------ Subject: Re: Hopfield Networks -- How Best to Run? From: usenet@nlm-mcs.arpa (usenet news poster) Organization: National Library of Medicine, Bethesda, Md. Date: 29 Mar 90 06:09:03 +0000 >(with simulated annealing, it's really the Boltzmann machine) I agree. > Now, there comes a stage in the annealing schedule >where jumps to lower enegry states occur much more frequently than >jumps to high energy states, and this ``transition'' period, which >bears resemblances to state transitions in solids/liquids, should be >held for as long as possible... But don't neglect a period or equilibration at the outset where the up and down transition occur with near equal probability. If you start with too low an initial temperature (or a bad starting point) most of the jumps will be downward, but you are really just using Monte Carlo as an inefficient minimization routine. A second point, since your "energy" function is easily differentiable you might consider computing "forces" and running dynamics rather than doing Monte Carlo. The efficiency of Monte Carlo depends critically on a good method for choosing new jumps successfully. Dynamics is, in a sense, mindless. You just follow the laws of motion. Annealing is achieved by gradully extracting kinetic "energy" from the system. >The best nets will have spent most time with the temperature >in the range a-b on the graph (deleted for brevity), where the >energy is decreasing quickest. But, if you don't have some detailed knowledge about the global surface, you could end up focusing on a local transition. >Has anyone *tried* a global calculation? This is a very good suggestion. If you can work on some scaled down nets, the reduced dimensionality of the problem will reduce the space you need to search enormously and allow global searches that might not otherwise be possible. The only way of knowing you have a global solution is that you always reach it from different starting conditions. What you can gain from reduced models is a feeling for the characteristics of the "energy" surface you are dealing with, and perhaps some guidance in attacking the full problem. David States Math is always the same, the difference between fields is who you think invented it. ------------------------------ Subject: neural nets that "count" From: harwood@cvl.umd.edu (David Harwood) Organization: Center for Automation Research, Univ. of Md. Date: 23 Apr 90 18:53:26 +0000 I recall reading something that surprised me - something reported about 15 years ago (perhaps in Science, perhaps by neuroscientist Richard Thompson, although I'm not sure) - that a large fraction of cortical neurons in mammals (5% or some large fraction) - apparently "count" specific stimulus -class events, up to a small number (6-8 or so, as I recall). That is, a particular counter cell fires characteristically upon an Nth event of its stimulus-class, independent of inter-stimulus time-delays (up to a limit). The characteristic behavior was supposed to be reliable. The report was on untrained response, but we can't assume that previous "learning" was not sometimes involved. There are important differences in power of pattern recognition between so-called "counter" automata and counter-free automata, allowing unbounded counting which of course we don't see in neurons. But it occurs to me that this sort of apparently very pervasive neural phenomena should have profound implications for neural learning and control, especially learning of asynchronous, partially-ordered programs of behavior or patterns of stimulus (of small, but multi-class/-modal orders). This would seem to occur at even lowest (unconscious) levels of neural processing. I don't know if there's been further neurophysiological investigation or explanation of this. It was left as kind of puzzle in the article, as I recall. But if things are as they were reported then, it would seem to be very important for theories of neural nets, real and artificial - apparently something basic, making for great complexity and power of recognition and control, which has been ignored or overlooked. ------------------------------ Subject: applications in chemistry From: russo@caip.rutgers.edu (Mark F. Russo) Organization: Rutgers Univ., New Brunswick, N.J. Date: 24 Apr 90 13:24:55 +0000 I am interested in applications of Neural Networks in protein structure prediction, chemical reaction product prediction, drug interaction effects ... and the like. and I'm looking for references to applications in these and closely related area's. I am aware of the following references... [1] Borman, S. "Neural Network Applications In Chemistry Begin to Appear", April, 24, 1989 CE$N [2] Holly, L. and Karplus, M. "Protein secondary structure prediction with a neural network", Proc. Natl. Acad. Sci., Vol. 86, Jan, 1989 [3] Qian, N. and Sejnowski, T. J. Mol. Biol. Vol 202 No. 4, 1988. Any other references or clues would be greatly appreciated! Mark F. Russo (russo@caip.rutgers.edu) _____________________________________________________________________________ uucp: {from.anywhere}!rutgers!caip.rutgers.edu!russo arpa: russo@caip.rutgers.edu or uucp: {from.anywhere}!rutgers!petsd!cccspd!ccscg!markr ------------------------------ Subject: About Hopfield neural-net From: IO92040@MAINE.BITNET (Dimi) Organization: University of Maine System Date: 01 May 90 01:12:09 +0000 Hello neural-nets fellows, I am somehow puzzled about one thing concerning the Hopefield learning scheme. As you know each memory state(pattern) is supposed to be stored in what so called memory matrix.If you want to reveal the correct memory state all you do is to get the inner product of the memory matrix and in the input vector which may be distorted(distorted memory state). Well, for one memory state you need at least an 8th dimensional memory matrix. For 2 memory states you need 16th dimensional matrix, according to the rule M=.15N. My question is if you want to create a memory matrix to store more than one memory states how do you really do it??How does your input vector looks like?Does it contain the memory states you want to reveal,sequentially?Is your memory matrix a combination of memory matrices that represent each memory state that you want it to contain or what??I would appreciate if somebody out there who understands more about Hopfield nets could be able to answer my question(if you understand what I am asking, I hope). Thanx a lot Dimi ------------------------------ Subject: Re: About Hopfield neural-net From: smagt@samantha.fwi.uva.nl (Patrick van der Smagt) Organization: FWI, University of Amsterdam Date: 04 May 90 07:05:35 +0000 There is a lot to be answered. First, the M=.15N looks better than it is. In a +1/-1 Hopfield network, storing a pattern means storing its inverse, too (by approximation, this is also true for the 0/1 network). So the number of actually different patterns you can effectively store is half of 0.15N. Note that Hopfield finds the 15% for a large number of neurons, so you shouldn't apply this to store one or two patterns. Furthermore, he measures the network's success according to the recall. When the #neurons:#patterns ratio exceeds 15%, recall errors are very severe. My experiments confirmed this. However, you can actually store as many as N patterns in an N-neuron network. How this is done is explained in %A A. D. Bruce %A A. Canning %A B. Forrest %A E. Gardner %A D. J. Wallace %B AIP Conference Proceedings 151, Neural Networks for Computing, %B Snowbird Utah, AIP %D 1986 %E J. S. Denker %T Learning and memory properties in fully connected networks %P 65--70 I'm presenting a paper on this at the Third International Conference on Industrial and Engineering Applications of Artificial Intelligence and Expert Systems (July 15-18, Charleston, SC). Patrick van der Smagt /\/\ \ / Organization: Faculty of Mathematics & Computer Science / \ University of Amsterdam, Kruislaan 409, _ \/\/ _ NL-1098 SJ Amsterdam, The Netherlands | | | | Phone: +31 20 525 7466 | | /\/\ | | Telex: 10262 hef nl | | \ / | | Fax: +31 20 592 5155 | | / \ | | email: smagt@fwi.uva.nl | | \/\/ | | | \______/ | \________/ /\/\ \ / / \ \/\/ ------------------------------ Subject: Library Circulation of Neural Network Journals From: will@ida.org (Craig Will) Date: Mon, 04 Jun 90 13:32:44 -0400 Circulation of Neural Network Journals in OCLC Libraries Journal Year Started Circulation IEEE Trans. on Neural Nets 1990 130 Neural Networks 1988 115 Neural Computation 1989 29 J. of Neural Network Comput 1989 21 Connection Science 1989 9 Network 1990 5 Inter. J. of Neural Networks 1989 4 Based on a search June 4, 1990 using the OCLC database of about 10,000 member libraries in the United States. Reli- able figures for Neural Network Review aren't available, apparently because of the change of publisher. No records were found for the International Journal of Neurocomputing. Craig A. Will will@ida.org Institute for Defense Analyses ------------------------------ Subject: A GA Tutorial and a GA Short Course From: "Dave Goldberg (dgoldber@ua1vm.ua.edu)" <DGOLDBER@UA1VM.ua.edu> Date: Thu, 07 Jun 90 16:02:23 -0500 A tutorial entitled "Genetic Algorithms and Classifier Systems" will be presented on Wednesday afternoon, August 1, at the AAAI conference in Boston, MA by David E. Goldberg (Alabama) and John R. Koza (Stanford). The course will survey GA mechanics, power, applications, and advances together with similar information regarding classifier systems and other genetics-based machine learning systems. For further information regarding this tutorial write to AAAI-90, Burgess Drive, Menlo Park, CA 94025, (415)328-3123. A five-day short course entitled "Genetic Algorithms in Search, Optimization, and Machine Learning" will be presented at Stanford University's Western Institute in Computer Science on August 6-10 by David E. Goldberg (Alabama) and John R. Koza (Stanford). The course presents in-depth coverage of GA mechanics, theory and application in search, optimization, and machine learning. Students will be encouraged to solve their own problems in hands-on computer workshops monitored by the course instructors. For further information regarding this course contact Joleen Barnhill, Western Institute in Computer Science, PO Box 1238, Magalia, CA 95954, (916)873-0576. ------------------------------ Subject: Quantum devices for Neural Nets? From: FOO@EVAX0.ENG.FSU.EDU Date: Fri, 08 Jun 90 09:51:43 -0400 I'd appreciate if anybody could give me pointers to info on quantum-well devices for implementing neural nets (i.e., atomic-level computing). My e-mail address is: foo@evax0.eng.fsu.edu I'll post a summary of all responses I received. Thank you. --simon foo dept. of electrical engineering florida state university tallahassee, fl 32306 ------------------------------ Subject: Special Issue Announcement From: Alex.Waibel@SPEECH2.CS.CMU.EDU Date: Sun, 10 Jun 90 20:09:38 -0400 ANNOUNCEMENT MACHINE LEARNING will be publishing a special issue devoted to connectionist models under the title: "Structured Connectionist Learning Systems: Methods and Real World Applications" MACHINE LEARNING publishes articles on all aspects of Machine Learning, and on occasion runs special issues on particular subtopics of special interest. This issue of the journal will emphasize conectionist learning systems that aim at real world applications. Papers are solicited on this topic. Five copies of the manuscript should be sent by August 3, 1990 to: Dr. Alex Waibel School of Computer Science Carnegie Mellon University Pittsburgh, PA 15213 Telephone: (412) 268-7676 Papers will be subject to the standard review process. ------------------------------ Subject: Re: Neuron Digest V6 #21 From: <OOMIDVAR%UDCVAX.BITNET@CORNELLC.cit.cornell.edu> Date: Mon, 11 Jun 90 18:45:00 -0400 I like to know if there is any actual working industrial inspection system based on neural networks technology? if so where and who is the technical person(Brain) in charge of it. 2. Is any one involved in parallel implementation of well known learning algorithms such as Backprop, LVQ, and Boltzmann machine? if so please let me know . I can also use technical papers in these areas. Thank you very much in advance and I will post the results of this . OOMIDVAR@UDC.BITNET ------------------------------ Subject: Another Request for NN Software From: SADLER@ADELPHI-HDLSIG1.ARMY.MIL Organization: Harry Diamond Labs, Adelphi, MD. Date: 12 Jun 90 13:12:17 -0500 I'm looking for NN simulation software, HOWEVER we are not a UNIX house. Platforms of choice are: 1. Vax running VMS 2. IBM/PC 3. Macintosh I currently do most signal processing simulation work in MATLAB on the VAX. (MATLAB is just great by the way.) Source code would also be of interest. thanks!! brian sadler ------------------------------ Subject: Request for biological NN info. From: Garth S. Barbour <garth@wam.umd.edu> Date: Wed, 13 Jun 90 10:36:40 -0400 Hi folks. I'm writing a paper on the relationship between natural and artificial neural networks as an independent study project for the summer. In particular, I will be looking for historical discoveries in the study of biological neural networks and how they did or did not influence the design of artificial neural networks. If anyone knows of any references which may be of use, I would greatly appreciate hearing about it. I am especially looking for any references on biological neural networks. I will publish a summary of the references recieved. My e-mail address is garth@cscwam.umd.edu. Thanks, Garth Barbour ------------------------------ Subject: Cognitive Science Society Meeting From: Stevan Harnad <harnad@clarity.Princeton.EDU> Date: Tue, 12 Jun 90 00:11:15 -0400 The XII Annual Conference of the Cognitive Science Society will take place at MIT, July 25-28, 1990. (Immediately preceding the meeting of the AAAI, also to take place in the Boston area). Conference Chair: M. Piattelli-Palmarini (MIT) Scientific Advisors: Beth Adelson (Tufts), Stephen M. Kosslyn (Harvard) Steven Pinker (MIT), Kenneth Wexler (MIT) Registration fees: Members 150$ (before July 1), $200 after July 1 non-members 185 225 student 90 110 Contact the MIT Conference Services, MIT Room 7- 111 Cambridge, MA 02139 Tel. (617) 253-1700 _______________________________________________ Outline of the program Tuesday July 24, Wednesday July 25 Tutorials: "Cognitive Aspects of Linguistic Theory", "Logic and Computability", Cognitive Neuroscience" (Require separate registrations) Wednesday, July 26 4.00 - 7.30 pm Registration at Kresge Auditorium 7.30 - 9.00 First plenary session : Kresge Main Auditorium Welcoming address by Samuel Jay Keyser, Assistant Provost of MIT, Co-Director of the MIT Center for Cognitive Science; Welcoming address by David E. Rumelhart (Stanford), Chairman of the Board of the Cognitive Science Society Keynote speaker: Noam Chomsky (MIT) "Language and Cognition" __________ Thursday, July 26 9.00 am - 11.15 am Symposia: Execution-Time Response: Applying Plans in a Dynamic World Kristian J. Hammond (University of Chicago), Chair Phil Agre (University of Chicago) Richard Alterman (Brandeis University) Reid Simmons (Carnegie Mellon University) R. James Firby (NASA Jet Propulsion Lab) Cognitive Aspects of Linguistic Theory Howard Lasnik (University of Connecticut), Chair David Pesetsky (Massachusetts Institute of Technology), Chair James T. Higginbotham (Massachusetts Institute of Technology) John McCarthy (University of Massachusetts) Perception, Computation and Categorization Whitman Richards (Massachusetts Institute of Technology), Chair Aaron Bobick (SRI International) Ken Nakayama (Harvard University) Allan Jepson (University of Toronto) Paper Presentations: Rule-Based Reasoning,Explanation and Problem-Solving Reasoning II: Planning 11.30 - 12.45 Plenary session Kresge Main Auditorium Keynote Speaker: Morris Halle (MIT) "Words and their Parts" Chair: Kenneth Wexler (MIT) __________________ Thursday, July 26 Afternoon 2.00 pm - 4.15 pm Symposia: Principle-Based Parsing Robert C. Berwick (Massachusetts Institute of Technology), Chair Steven P. Abney (Bell Communications Research) Bonnie J. Dorr (Massachusetts Institute of Technology) Sandiway Fong (Massachusetts Institute of Technology) Mark Johnson (Brown University) Edward P. Stabler, Jr. (University of California, Los Angeles) Recent Results in Formal Learning Theory Kevin T. Kelly (Carnegie Mellon University) Clark Glymour (Carnegie Mellon University), Chair Self-Organizing Cognitive and Neural Systems Stephen Grossberg (Boston University), Chair Ennio Mingolla (Boston University) Michael Rudd (Boston University) Daniel Bullock (Boston University) Gail A. Carpenter (Boston University) Action Systems: Planning and Execution Emilio Bizzi (Massachusetts Institute of Technology), Chair Michael I. Jordan (Massachusetts Institute of Technology) Paper presentations Reasoning : Analogy Learning and Memory : Acquisition 4.30 - 5.45 Plenary Session (Kresge Main Auditorium) Keynote Speaker: Amos Tversky (Stanford) "Decision under conflict" Chair: Daniel N. Osherson (MIT) Banquet ______________ Friday July 27 9.00 - 11.45 am Symposia: What's New in Language Acquisition ? Steven Pinker and Kenneth Wexler (MIT), Chair Stephen Crain (University of Connecticut) Myrna Gopnik (McGill University) Alan Prince (Brandeis University) Michelle Hollander, John Kim, Gary Marcus, Sandeep Prasada, Michael Ullman (MIT) Attracting Attention Ann Treisman (University of California,Berkeley), Chair Patrick Cavanagh (Harvard University) Ken Nakayama (Harvard University) Jeremy M. Wolfe (Massachusetts Institute of Technology) Steven Yantis (Johns Hopkins University) A New Look at Decision Making Susan Chipman (Office of Naval Research) and Judith Orasanu (Army Research Institute and Princeton University),Chair Gary Klein (Klein Associates) John A. Swets (Bolt Beranek & Newman Laboratories) Paul Thagard (Princeton University) Marvin S. Cohen (Decision Science Consortium, Inc.) Designing an Integrated Architecture: The Prodigy View Jaime G. Carbonell (Carnegie Mellon University), Chair Yolanda Gil (Carnegie Mellon University) Robert Joseph (Carnegie Mellon University) Craig A. Knoblock (Carnegie Mellon University) Steve Minton (NASA Ames Research Center) Manuela M. Veloso (Carnegie Mellon University) Paper presentations: Reasoning : Categories and Concepts Language : Pragmatics and Communication 11.30 - 12. 45 Plenary Session (Kresge main Auditorium) Keynote speaker: Margaret Livingstone (Harvard) "Parallel Processing of Form, Color and Depth" Chair: Richard M. Held (MIT) _________________________ Friday, July 27 afternoon 2.00 - 4.15 pm Symposia: What is Cognitive Neuroscience? David Caplan (Harvard Medical School) and Stephen M. Kosslyn (Harvard University), Chair Michael S. Gazzaniga (Dartmouth Medical School) Michael I. Posner (University of Oregon) Larry Squire (University of California, San Diego) Computational Models of Category Learning Pat Langley (NASA Ames Research Center) and Michael Pazzani (University of California, Irvine), Chair Dorrit Billman (Georgia Institute of Technology) Douglas Fisher (Vanderbilt University) Mark Gluck (Stanford University) The Study of Expertise: Prospects and Limits Anders Ericsson (University of Colorado, Boulder),Chair Neil Charness (University of Waterloo) Vimla L. Patel and Guy Groen (McGill University) Yuichiro Anzai (Keio University) Fran Allard and Jan Starkes (University of Waterloo) Keith Holyoak (University of California, Los Angeles), Discussant Paper presentations: Language (Panel 1) : Phonology Language (Panel 2) : Syntax 4.30 - 5.45 Keynote speaker: Anne Treisman (UC Berkeley) "Features and Objects" Chair: Stephen M. Kosslyn (Harvard) Poster Sessions I. Connectionist Models II. Machine Simulations and Algorithms III. Knowledge and Problem-Solving __________________________________ Saturday, July 28 9.00 am - 11.15 am Symposia: SOAR as a Unified Theory of Cognition: Spring 1990 Allen Newell (Carnegie Mellon University), Chair Richard L. Lewis (Carnegie Mellon University) Scott B. Huffman (University of Michigan) Bonnie E. John (Carnegie Mellon University) John E. Laird (University of Michigan) Jill Fain Lehman (Carnegie Mellon University) Paul S. Rosenbloom (University of Southern California) Tony Simon (Carnegie Mellon University) Shirley G. Tessler (Carnegie Mellon University) Neonate Cognition Richard Held (Massachusetts Institute of Technology), Chair Jane Gwiazda (Massachusetts Institute of Technology) Renee Baillargeon (University of Illinois) Adele Diamond (University of Pennsylvania) Jacques Mehler (CNRS, Paris, France) Discussant Conceptual Coherence in Text and Discourse Arthur C. Grasser (Memphis State University), Chair Richard Alterman (Brandeis University) Kathleen Dahlgren (Intelligent Text Processing, Inc.) Bruce K. Britton (University of Georgia) Paul van den Broek (University of Minnesota) Charles R. Fletcher (University of Minnesota) Roger J. Kreuz (Memphis State University) Richard M. Roberts (Memphis State University) Tom Trabasso and Nancy Stein Paper presentations: Causality,Induction and Decision-Making Vision (Panel 1) : Objects and Features Vision (Panel 2) : Imagery Language : Lexical Semantics Case-Based Reasoning 11.30 - 12.45 Keynote Speaker Ellen Markman (Stanford) "Constraints Children Place on Possible Word Meanings" Chair: Susan Carey (MIT) Lunch presentation: "Cognitive Science in Europe: A Panorama" Chair: Willem Levelt (Max Planck, Nijmegen). Informal presentations by: Jacques Mehler (CNRS, Paris), Paolo Viviani (University of Geneva), Paolo Legrenzi (University of Trieste), Karl Wender (University of Trier). _____________________________ Saturday 28 Afternoon 2.00 - 3.00 pm Paper presentations: Vision : Attention Language Processing Educational Methods Learning and Memory Agents, Goals and Constraints 3.15 - 4.30 Keynote Speaker: Roger Schank (Norhwestern) "The Story is the Message: Memory and Instruction" Chair: Beth Adelson (Tufts) 4.30 - 5.45 Keynote Speaker: Stephen Jay Gould (Harvard) "Evolution and Cognition" Chair: Steven Pinker (MIT) ------------------------------ End of Neuron Digest [Volume 6 Issue 40] ****************************************
dsikka@td2cad.intel.com (Digvijay Sikka) (06/14/90)
Hi Folks: Several months ago there was a posting on the net regarding a paper that attempted to study relationships between Bayesian (Probabilistic) Networks, and Neural Networks. Since I am interested in such a study, I would appreciate if someone out there can either repost it or send it me via e-mail. In case if it is not available, I would appreciate any pointers in this regard. Thanx, Digvijay.
mgv@usceast.UUCP (Marco Valtorta) (06/15/90)
In article <3142@td2cad.intel.com> dsikka@aries.intel.com (Digvijay Sikka) writes: >Hi Folks: > Several months ago there was a posting on the net regarding a >paper that attempted to study relationships between Bayesian >(Probabilistic) Networks, and Neural Networks. Since I am interested in >such a study, I would appreciate if someone out there can either repost >it or send it me via e-mail. In case if it is not available, I would >appreciate any pointers in this regard. I have some comments on that in a paper in the Proceedings of the Seventh International Conference on Machine Learning. The Proceedings are to be published by Morgan Kaufmann. The conference will take place next week. If you are going there, look out for me! Apart from presenting the paper, I am organizing a discussion group on knowledge base refinement. I have a couple of other papers that address the topic you are interested in. One is going to be published by the *International Journal of Approximate Reasoning*, probably in January 1991. The other is somewhere (:-)) in the submission/revision process. From the algorithmic standpoint, there are many similarities between knowledge base refinement and neural network training. My dissertation was on the topic of knowledge base refinement. At the time I started that work, neural networks were a dead field. Now, the balance of interest has shifted in the opposite direction! > >Thanx, > >Digvijay. You are welcome! Marco Valtorta usenet: ...!ncrcae!usceast!mgv Department of Computer Science internet: mgv@cs.scarolina.edu University of South Carolina tel.: (1)(803)777-4641 Columbia, SC 29208 tlx: 805038 USC U.S.A. fax: (1)(803)777-3065 usenet from Europe: ...!mcvax!uunet!ncrlnk!ncrcae!usceast!mgv