Received: from BUACCA by BUACCA.BU.EDU (Mailer R2.08 PTF009) with BSMTP id 5154; Sat, 07 Nov 92 15:42:11 EST Received: from noc2.dccs.upenn.edu by BUACCA.BU.EDU (IBM VM SMTP R1.2.1) with TCP; Sat, 07 Nov 92 15:42:08 EST Received: from CATTELL.PSYCH.UPENN.EDU by noc2.dccs.upenn.edu id AA28469; Sat, 7 Nov 92 15:38:25 -0500 Return-Path: Received: from LOCALHOST by cattell.psych.upenn.edu id AA02176; Sat, 7 Nov 92 15:06:30 EST Posted-Date: Sat, 07 Nov 92 15:06:01 EST From: "Neuron-Digest Moderator" To: Neuron-Distribution:; Subject: Neuron Digest V10 #16 (discussion + jobs + software) Reply-To: "Neuron-Request" X-Errors-To: "Neuron-Request" Organization: University of Pennsylvania Date: Sat, 07 Nov 92 15:06:01 EST Message-Id: <2165.721166761@cattell.psych.upenn.edu> Sender: marvit@cattell.psych.upenn.edu Neuron Digest Saturday, 7 Nov 1992 Volume 10 : Issue 16 Today's Topics: educational stock market game Clinical and Research Positions at the VA GRECC(Salt Lake City) position offer from RIKEN "Help!!" - protein structures and fault diagnosis Neural Nets and Visual tracking? faculty positions (U Mass) faculty position (Bristol, UK) Connectionist Models Summer School 1993 New version of Self-Organizing Maps PD program package Send submissions, questions, address maintenance, and requests for old issues to "neuron-request@cattell.psych.upenn.edu". The ftp archives are available from cattell.psych.upenn.edu (130.91.68.31). Back issues requested by mail will eventually be sent, but may take a while. ---------------------------------------------------------------------- Subject: educational stock market game From: Lawrence Davenport III Date: Sat, 24 Oct 92 01:40:27 -0500 Dear "Neuron-Digest: If you have an interest in the stock market, want to learn about investment, want to receive some investment newsletters or want to try out your investment skills, then please read on. Otherwise, please hit the delete key now. 1. THE EVENT - WALL STREET INVESTMENT CHALLENGE: The Wall Street Investment Challenge is a contest in which each participant manages a fictional account of $1,000,000 in the stock market. During the 3 months contest, each participants will receive various top-rated investment newsletters. The winners of the contest will receive a set of prizes ranging from $25,000 to $100. 2. THE EDUCATIONAL ASPECT In addition to the fun and excitement that are associated with an investment challenge, the Wall Street Investment Challenge has created a three facet approach to help the participants to improve their investment skills. A). Learning Of The Investment Principles As a participant, you will receive 3 monthly issues of the Wall Street Investment Review which reports the latest research findings in the nation's best business schools on the subject of investment approaches that have yielded high profit or low risk. All investment concepts are explained in simple to understand language so that the readers can gain immediate understanding and benefits. B). Transforming The Investment Principles To Practical Stock Selection You will also receive highly regarded investment newsletters, including 6 issues of Dow Theory Forecasts and 6 issues of The Cabot Market Letter. Both of these two investment newsletters have been rated in the top 10, out of hundreds of newsletters, in Hubert's Guide To The Investment Newsletters. You can improve your ability to apply your investment knowledge to actual stock selection by observing the practices of these highly successful investment advisors. C). Sharpen The Investment Instincts In The Real-Time Financial World Many day to day financial and political developments have real impacts on short term and long term stock investment. Hence, you can sharpen your investment instincts by actively managing one or more portfolios in the Wall Street Investment Challenge. 3. THE PRIZES National Prizes: The 20 portfolios with the highest balance at the end of the competition will win the following cash prizes: 1st - 25,000 2nd - 4,000 3rd - 2,500 4th - 2,000 5th - 1,500 6th thru 20th - 1,000 each Division Champion Each participant is assigned to a division. Each division has 100 participants. At the end of the competition, the participant with the highest portfolio value in the division will win the title of Division Champion and will receive a $500 prize. League Champion If you and your friends register together with at least 10 portfolios, you can form a league with a name of your choice. The best trader in your league will win the title of the League Champion and will receive a cash prize of $100. 4. OTHER INFO Cost - $49.95 entry fee Duration of the contest - from 11/1/92 to 1/31/93 Allowable trades - buy, sell and short all NYSE, AMEX and NASDAQ (OTC) stocks To register or for more info, please call 1 (800) 964-6463, 24 hours a day, 7 days a week. ------------------------------ Subject: Clinical and Research Positions at the VA GRECC(Salt Lake City) From: soller%asylum.cs.utah.edu@cs.utah.edu (Jerome Soller) Organization: University of Utah CS Dept Date: 25 Oct 92 00:55:44 +0000 I thought the following job posting would be of interest to researchers in the neural network, biological models of aging, and intelligent medical computing systems area. Some examples of research projects at the VA GRECC in these areas include: applications of neural networks to large database systems, application of neural networks to prediction of heart disease(Charles Rosenberg), integrated data collection environments using pen based computers(Steven Fehlauer), Semantic Networks(Judith Graves), expert systems, etc.., to real problems in medicine and nursing. Close ties exist between the GRECC and U. of Utah Departments of Internal Medicine, Bioengineering, Computer Science, Psychology, Medical Informatics, and Nursing Informatics. Another separate VA center in Utah(VA Regional Information Systems Center, one of 7 in the U.S.) provides access to the VA's DHCP database, which is a standard database for 172 VA hospitals. Sincerely, Jerome Soller VA GRECC Ph.D. Candidate, U. of Utah Department of Computer Science soller@cs.utah.edu - ------------------------------------------------------------------------- The Salt Lake City Geriatric Research, Education and Clinical Center (GRECC) and the University of Utah School of Medicine are recruiting individuals to join the VA/University program in Geriatric Internal Medicine. University faculty rank is dependent on qualifications. Candidates must hold an M.D. and/or Ph.D. and have interest in one or more of the following: Clinical Geriatric Internal Medicine Basic Biological Mechanisms of Aging Computer-based Strategies for "Real-Time" Assistance in the Delivery of Health Care Appointments will be in the SLC VAMC GRECC and in the appropriate University of Utah academic department. Send curriculum vitae and bibliography to: Gerald Rothstein, M.D. SLC GRECC (182) 500 Foothill Blvd. SLC, UT 84148 Closing date March 31, 1993 or until suitable candidates are identifie d. Please call 801-582-1565 ext 4161 for further information. The Department of Veterans Affairs and the University of Utah are AA/EEO Employers. C. Steven Fehlauer, M.D. GRECC Research Investigator Assistant Professor of Medicine University of Utah Charles Rosenberg, Ph.D. GRECC Research Investigator ------------------------------ Subject: position offer from RIKEN From: Takayuki Ito Date: Mon, 26 Oct 92 10:38:11 +0200 I am Ito in NHK(Japan Broadcasting Corporation). Dr. Tanaka in RIKEN Institute asked me to post this letter. For more details, please contact with him by telephone or fax. - ------------------------------------ Offer of a position for public subscription RIKEN Institute, Information Science Laboratory Researcher Field: Physiological, anatomical, and psychological studies of higher brain functions, and development of related methodology. Available from April 1, 1993 Condition: Ph.D or scheduled by April 1, 1993. No elder than 34 on February 1, 1993. Any nationality. Inquires to Dr. Keiji Tanaka, Chief of Information Science Laboratory, fax: +81-48-462-4696, tel: +81-48-462-1111 ext.6411 - ------------------------------------ ---------------------------------------------- Takayuki Ito (itot@strl.nhk.or.jp) NHK Science and Technical Research Labs. 1-10-11, Kinuta, Setagaya-ku, Tokyo 157 Japan Tel.+81-3-5494-2369, Fax.+81-3-5494-2371 ---------------------------------------------- ------------------------------ Subject: "Help!!" - protein structures and fault diagnosis From: VEMURI@icdc.llnl.gov Date: Tue, 27 Oct 92 14:35:00 -0800 Help!! I am getting interested in looking into the applications of neural nets 1. to protein structure problems. 2. to fault diagnosis (in electromechanical systems) I have seen several times discussion about these two topics in these columns. At that time I wasn't interested. NOW I AM!! Can any one help with some bibliography material on these two topics. Thanks you V. Vemuri Dept. of Applied Science University of California Livermore, CA 94550 "Vemuri@icdc.llnl.gov" ------------------------------ Subject: Neural Nets and Visual tracking? From: Denis Mareschal Date: Wed, 28 Oct 92 15:20:47 +0000 Hi, I'm looking for references on the application of Neural-networks to visual-tracking. In particular, the ability to predict or anticipate futur positions based on information about the current history of the trajectory as well as THE DEVELOPMENT of this ability. I'm not very familiar with the vision literature but a preliminary search hasn't turned anything up. I am aware of NN applications to time-series analysis but I would prefer to find work dealing more explicitly with vision. Any help would be greatly appreciated and of course a list of references will be compiled and posted if sufficient requests are made. Thanks a lot Cheers, Denis Mareschal Department of Psychology Oxford maresch@ox.ac.uk ------------------------------ Subject: faculty positions (U Mass) From: Andy Barto Date: Fri, 30 Oct 92 18:53:42 -0500 UNIVERSITY OF MASSACHUSETTS AMHERST Faculty and Research Scientist Positions The Department of Computer Science invites applications for one-three tenure-track faculty positions at the assistant and associate levels and several research-track faculty and postdoctoral positions at all levels, in all areas of computer science. Applicants should have a Ph.D. in computer science or related area and should show evidence of exceptional research promise. Senior level candidates should have a record of distinguished research. Salary is commensurate with education and experience. Our Department has grown substantially over the past five years and currently has 30 tenure-track faculty and 8 research faculty, approximately 10 postdoctoral research scientists, and 160 graduate students. Continued growth is expected over the next five years. We have ongoing research projects in robotics, vision, natural language processing, expert systems, distributed problem solving, machine learning, artificial neural networks, person-machine interfaces, distributed processing, database systems, information retrieval, operating systems, object-oriented systems, persistent object management, real-time systems, real-time software development and analysis, programming languages, computer networks, theory of computation, office automation, parallel computation, computer architecture, and medical informatics (with the UMass Medical School). Send vita, along with the names of four references to Chair of Faculty Recruiting, Department of Computer Science, University of Massachusetts, Lederle Graduate Research Center, Amherst, MA 01003 by February 1, 1993 (or Email inquiries can be sent to facrec@cs.umass.edu). An Affirmative Action/Equal Opportunity Employer ------------------------------ Subject: faculty position (Bristol, UK) From: I C G Campbell Date: 02 Nov 92 15:04:01 +0000 FACULTY POSITION UNIVERSITY OF BRISTOL, UNITED KINGDOM Department of Computer Science Applications are invited for a Lectureship in Computer Science now tenable. FURTHER PARTICULARS The Department is part of the Faculty of Engineering. It has a complement of eighteen full-time UFC-funded staff members, together with a further twelve full-time outside-funded staff, and two Visiting Industrial Professors: Professor J. M. Taylor (Director, Hewlett-Packard Research Laboratories, Bristol) and Professor I. M. Barron. There are three Professors in the Department: Professor M. H. Rogers, who is Head of Department, and Professors J. W. Lloyd and D. H. D. Warren. The Department has substantial research funding from ESPRIT, SERC, industry and government. The Department concentrates its research in three main areas: Logic Programming Parallel Computing Machine Intelligence although a number of other topics are also being pursued. For this appointment, we are looking for a strong candidate in any area of Computer Science, although preference will be given to candidates with research interests in Parallel Computing or Machine Intelligence. We are particularly looking for candidates whose interests will broaden and complement our current work in these areas. Current work in Parallel Computing covers a range of areas, including parallel logic programming systems and languages, memory organisation for multiprocessor architectures, shared data models for transputer-based systems, and parallel applications especially for engineering problems and computer graphics. We are seeking to broaden and strengthen this research. Candidates with a strong background in computer architecture would be particularly welcome. Current work in Machine Intelligence centres mainly on Computer Vision and Speech Processing. One major project in Computer Vision is the development of an autonomous road vehicle, based on real-time image analysis. Other research projects in Computer Vision include vehicle number plate decoding, aircraft engine inspection, and visual flow monitoring. Current work on Speech Processing within the Department concentrates on speech synthesis, but the Faculty supports a Centre for Communications Research within which there is a Speech Research Group incorporating researchers in most aspects of speech technology, including speech recognition, speech coding, speech perception, and the design of speech interfaces. There is an interest in neural network theory and neural computing elsewhere in the Faculty and we would welcome applications from candidates in this area. The Department has a flourishing undergraduate and post-graduate teaching programme and participates in degree programmes in the Engineering, Science and Arts Faculties. These programmes lead to B.Sc. degrees in Computer Science, and Computer Science with Mathematics, a B.Eng. in Computer Systems Engineering, a B.A. in Computer Science and a Modern Language, and M.Sc. degrees in Computer Science, Foundations of Artificial Intelligence, and Information Engineering. The salary will be within the Lecturer Scale and the initial placement will depend on age, qualifications and experience. The closing date for applications is 27th November 1992. Further particulars may be obtained from the Head of the Computer Science Department (tel: 0272-303584; or e-mail: barbara@bristol.uk.ac.compsci). ------------------------------ Subject: Connectionist Models Summer School 1993 From: "Michael C. Mozer" Date: Mon, 02 Nov 92 14:35:14 -0700 CALL FOR APPLICATIONS CONNECTIONIST MODELS SUMMER SCHOOL University of Colorado Boulder, Colorado June 21 - July 3, 1993 The University of Colorado will host the 1993 Connectionist Models Summer School from June 21 to July 3, 1993. The purpose of the summer school is to provide training to promising young researchers in connectionism (neural networks) by leaders of the field and to foster interdisciplinary collaboration. This will be the fourth such program in a series that was held at Carnegie-Mellon in 1986 and 1988 and at UC San Diego in 1990. Previous summer schools have been extremely successful and we look forward to the 1993 session with anticipation of another exciting event. The summer school will offer courses in many areas of connectionist modeling, with emphasis on artificial intelligence, cognitive science, cognitive neuroscience, theoretical foundations, and computational methods. Visiting faculty (see list of invited faculty below) will present daily lectures and tutorials, coordinate informal workshops, and lead small discussion groups. The summer school schedule is designed to allow for significant interaction among students and faculty. As in previous years, a proceedings of the summer school will be published. Applications will be considered only from graduate students currently enrolled in Ph.D. programs. About 50 students will be accepted. Admission is on a competitive basis. Tuition will be covered for all students, and we expect to have scholarships available to subsidize housing and meal costs, which will run approximately $300. Applications should include the following materials: * a one-page statement of purpose, explaining major areas of interest and prior background in connectionist modeling and neural networks; * a vita, including academic history, publications (if any), and a list of relevant courses taken with instructors' names and grades received; * two letters of recommendation from individuals familiar with the applicants' work; and * if room and board support is requested, a statement from the applicant describing potential sources of financial support available (department, advisor, etc.) and the estimated extent of need. We hope to have sufficient scholarship funds available to provide room and board to all accepted students regardless of financial need. Applications should be sent to: Connectionist Models Summer School c/o Institute of Cognitive Science Campus Box 344 University of Colorado Boulder, CO 80309 All application materials must be received by March 1, 1993. Decisions about acceptance and scholarship awards will be announced April 15. If you have additional questions, please write to the address above or send e-mail to "cmss@cs.colorado.edu". Organizing Committee Jeff Elman (UC San Diego) Mike Mozer (University of Colorado) Paul Smolensky (University of Colorado) Dave Touretzky (Carnegie-Mellon) Andreas Weigend (Xerox PARC and University of Colorado) Additional faculty will include: Andy Barto (University of Massachusetts, Amherst) Gail Carpenter (Boston University) Jack Cowan (University of Chicago) David Haussler (UC Santa Cruz) Geoff Hinton (University of Toronto) Mike Jordan (MIT) John Kruschke (Indiana University) Jay McClelland (Carnegie-Mellon) Steve Nowlan (Salk Institute) Dave Plaut (Carnegie-Mellon) Jordan Pollack (Ohio State) Dave Rumelhart (Stanford) Terry Sejnowski (UC San Diego and Salk Institute) ------------------------------ Subject: New version of Self-Organizing Maps PD program package From: lvq@cochlea.hut.fi (LVQ_PAK) Date: Tue, 03 Nov 92 12:30:27 +0700 ************************************************************************ * * * SOM_PAK * * * * The * * * * Self-Organizing Map * * * * Program Package * * * * Version 1.2 (November 2, 1992) * * * * Prepared by the * * SOM Programming Team of the * * Helsinki University of Technology * * Laboratory of Computer and Information Science * * Rakentajanaukio 2 C, SF-02150 Espoo * * FINLAND * * * * Copyright (c) 1992 * * * ************************************************************************ Some time ago we released the software package "LVQ_PAK" for the easy application of Learning Vector Quantization algorithms. Corresponding public-domain programs for the Self-Organizing Map (SOM) algorithms are now available via anonymous FTP on the Internet. "What does the Self-Organizing Map mean?", you may ask --- See the following reference, then: Teuvo Kohonen. The self-organizing map. Proceedings of the IEEE, 78(9):1464-1480, 1990. In short, Self-Organizing Map (SOM) defines a 'non-linear projection' of the probability density function of the high-dimensional input data onto the two-dimensional display. SOM places a number of reference vectors into an input data space to approximate to its data set in an ordered fashion. This package contains all the programs necessary for the application of Self-Organizing Map algorithms in an arbitrary complex data visualization task. This code is distributed without charge on an "as is" basis. There is no warranty of any kind by the authors or by Helsinki University of Technology. In the implementation of the SOM programs we have tried to use as simple code as possible. Therefore the programs are supposed to compile in various machines without any specific modifications made on the code. All programs have been written in ANSI C. The programs are available in two archive formats, one for the UNIX-environment, the other for MS-DOS. Both archives contain exactly the same files. These files can be accessed via FTP as follows: 1. Create an FTP connection from wherever you are to machine "cochlea.hut.fi". The internet address of this machine is 130.233.168.48, for those who need it. 2. Log in as user "anonymous" with your own e-mail address as password. 3. Change remote directory to "/pub/som_pak". 4. At this point FTP should be able to get a listing of files in this directory with DIR and fetch the ones you want with GET. (The exact FTP commands you use depend on your local FTP program.) Remember to use the binary transfer mode for compressed files. The som_pak program package includes the following files: - Documentation: README short description of the package and installation instructions som_doc.ps documentation in (c) PostScript format som_doc.ps.Z same as above but compressed som_doc.txt documentation in ASCII format - Source file archives (which contain the documentation, too): som_p1r2.exe Self-extracting MS-DOS archive file som_pak-1.2.tar UNIX tape archive file som_pak-1.2.tar.Z same as above but compressed An example of FTP access is given below unix> ftp cochlea.hut.fi (or 130.233.168.48) Name: anonymous Password: ftp> cd /pub/som_pak ftp> binary ftp> get som_pak-1.2.tar.Z ftp> quit unix> uncompress som_pak-1.2.tar.Z unix> tar xvfo som_pak-1.2.tar See file README for further installation instructions. All comments concerning this package should be addressed to som@cochlea.hut.fi. ************************************************************************ ------------------------------ End of Neuron Digest [Volume 10 Issue 16] ***************************************** Received: from BUACCA by BUACCA.BU.EDU (Mailer R2.08 PTF009) with BSMTP id 0849; Sun, 08 Nov 92 17:51:45 EST Received: from noc2.dccs.upenn.edu by BUACCA.BU.EDU (IBM VM SMTP R1.2.1) with TCP; Sun, 08 Nov 92 17:51:42 EST Received: from CATTELL.PSYCH.UPENN.EDU by noc2.dccs.upenn.edu id AA09041; Sun, 8 Nov 92 17:46:40 -0500 Return-Path: Received: from LOCALHOST by cattell.psych.upenn.edu id AA15784; Sun, 8 Nov 92 17:07:08 EST Posted-Date: Sun, 08 Nov 92 17:06:40 EST From: "Neuron-Digest Moderator" To: Neuron-Distribution:; Subject: Neuron Digest V10 #17 Reply-To: "Neuron-Request" X-Errors-To: "Neuron-Request" Organization: University of Pennsylvania Date: Sun, 08 Nov 92 17:06:40 EST Message-Id: <15781.721260400@cattell.psych.upenn.edu> Sender: marvit@cattell.psych.upenn.edu Neuron Digest Sunday, 8 Nov 1992 Volume 10 : Issue 17 Today's Topics: Stock market scam more like it Fellowship Position: Neural Computation in Neurology Clinical and Research Opportunities at the VA GRECC(Salt Lake City) Lectures announcement (Belgium) New version of Self-Organizing Maps PD program package Audio Synthesizer NIPS*92 and CME travel Re: Hotel reservation deadline for NIPS workshops Send submissions, questions, address maintenance, and requests for old issues to "neuron-request@cattell.psych.upenn.edu". The ftp archives are available from cattell.psych.upenn.edu (130.91.68.31). Back issues requested by mail will eventually be sent, but may take a while. ---------------------------------------------------------------------- Subject: Stock market scam more like it From: bradski@cns.bu.edu (Gary Bradski) Date: Sat, 07 Nov 92 21:44:41 -0500 [[ See Editor's Note at the end. ]] I can't see how you let that "educational stock market game" through. What if I send you something asking people for money for an "educational" poker game where I'd be willing to give part of my take (for running the game) to the winner? A scam is a scam and you've just helped facilitate one for 50 bucks a pop. This belongs in Atlantic City, not on the net. --Gary @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@ --------------- Gary Bradski I'net: bradski@park.bu.edu | reverberate | Cognitive and Neural Systems --------------- Boston University. | V V 111 Cummington St, Boston MA 02215 ^ Y 617/ 353-6426 ^ ^ | -------------- | or die! | @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@ -------------- [[ Editor's Note: Because of Gary's note, I reread last issues's submission of the stock market game and completly agree with him. I had, in fact, completely missed the entry fee portion of the message; if I had understood, I probably would not have accepted it for publication. I know of other "stock market games" which do not require entry fees and had assumed this was similar. As long-time readers know, I routinely reject only two types of articles: Jobs *wanted* (i.e., resumes) and blatantly commercial postings designed primarily for financial gain. In retrospect, the "stock market game" falls under the second category and violates the non-commercial intent of the Internet. I strongly recommend Neuron Digest readers *not* participate. -PM ]] ------------------------------ Subject: Fellowship Position: Neural Computation in Neurology From: "James A. Reggia" Date: Tue, 03 Nov 92 12:36:59 -0500 Research Training Fellowship in Neural Modelling Available (MD Degree required) The Clinical Stroke Research Center at the University of Maryland School of Medicine will offer two Junior Javits research fellowships starting July 1, 1993. One of these positions provides research training in the use of neural networks in cerebrovascular disease. A clinical back- ground (MD degree and specialization in neurology) is required. The Fellowship is for two years and is research intensive, but would also usually involve some clinical work in the Stroke Center. There is substantial flexibility in the details of the research training and research work. The first year salary is anticipated to be $33,000 plus fringe benefits. To apply send a letter and curriculum vitae to Dr. Thomas Price Director, Clinical Stroke Research Center University of Maryland Hospital 22 South Greene Street Baltimore, MD 21201 Questions about the research program can be sent to: Jim Reggia reggia@cs.umd.edu ------------------------------ Subject: Clinical and Research Opportunities at the VA GRECC(Salt Lake City) From: soller@asylum.cs.utah.edu (Jerome Soller) Date: Tue, 03 Nov 92 17:04:57 -0700 Dear Neuron Digest: Please post our job notice for the Research Investigator and clinical positions position at the VA GRECC(I apologize for the duplicate posting of this message to comp.ai.neural-nets). Some examples of computer research projects at the VA GRECC in these areas include: applications of neural networks to large database systems, application of neural networks to prediction of heart disease(Charles Rosenberg), integrated data collection environments using pen based computers(Steven Fehlauer), Semantic Networks(Judith Graves), expert systems(Bayesian and fuzzy logic varieties), etc.., to real problems in medicine and nursing. Close ties exist between the GRECC and U. of Utah Departments of Internal Medicine, Bioengineering, Computer Science, Psychology, Medical Informatics, and Nursing Informatics. Another separate VA center in Utah(VA Regional Information Systems Center, one of 7 in the U.S.) provides access to the VA's DHCP database, which is a standard database for 172 VA hospitals. Some examples of neurological work done in cooperation with our group are Dustman and Emerson's work on the effect of exercise on the cognition of elderly people. Sincerely, Jerome Soller Computer Research Engineer, VA GRECC Ph.D. Candidate, U. of Utah Department of Computer Science soller@cs.utah.edu The Salt Lake City Geriatric Research, Education and Clinical Center (GRECC) and the University of Utah School of Medicine are recruiting individuals to join the VA/University program in Geriatric Internal Medicine. University faculty rank is dependent on qualifications. Candidates must hold an M.D. and/or Ph.D. and have interest in one or more of the following: Clinical Geriatric Internal Medicine Basic Biological Mechanisms of Aging Computer-based Strategies for "Real-Time" Assistance in the Delivery of Health Care Appointments will be in the SLC VAMC GRECC and in the appropriate University of Utah academic department. Send curriculum vitae and bibliography to: Gerald Rothstein, M.D. SLC GRECC (182) 500 Foothill Blvd. SLC, UT 84148 Closing date March 31, 1993 or until suitable candidates are identified. Please call 801-582-1565 ext 4161 for further information. The Department of Veterans Affairs and the University of Utah are AA/EEO Employers. C. Steven Fehlauer, M.D. (fehlauer@msscc.med.utah.edu) GRECC Research Investigator Assistant Professor of Medicine University of Utah Charles Rosenberg, Ph.D. (crr@cogsci.psych.utah.edu) GRECC Research Investigator ------------------------------ Subject: Lectures announcement (Belgium) From: Laurence Leherte Date: Wed, 04 Nov 92 10:14:46 Announcement: A Series of Lectures in the field of Chemistry and Artificial Intelligence will be held at the Facultes Universitaires Notre-Dame de la Paix (Namur, Belgium), on November the 19th. ***************************** * MOLECULAR SCENE ANALYSIS * ***************************** Janice Glasgow, Dept. of Computing and Information Science Queen's University, Kingston, Canada Frank H. Allen, Cambridge Structural Data Centre, Cambridge, UK. Suzanne Fortier, Dept. of Chemistry, Queen's University, Queen's University, Kingston, Canada The concept of "scene analysis" has been used in the context of machine vision to refer to the set of processes associated with the classification and understanding of complex images. Such analyses rely on the availibility of domain knowledge in the form of structural templates, rules or heuristics to locate and identify features in a scene. By analogy we use the phrase "molecular scene analysis" to refer to the processes associated with the reconstruction and interpretation of molecular structures and molecular interactions. This presentation, in three parts, will describe some fundamental aspects of the Molecular Scene Analysis project. Computational Imagery : Janice Glasgow - --------------------------------------- At the core of our knowledge-based approach to molecular scene analysis is the concept of imagery, that is the ability to reason with three-dimensional images of molecular structure. To provide a computational framework that can imitate the human visualization abilities, we are designing image representations that make explicit the fundamental spatial and visual charactersitics of a molecular scene. Applying the computational reasoning techniques of molecular imagery to the information accumulated in a crystallographic knowledge base provides the "intelligence" vital to molecular scene analysis. >From Databases to Knowledge Bases : Frank Allen - ----------------------------------------------- Full 2D chemical and 3D crystallographic (coordinate) data for 180,000 crystal structures are now held in computerized databases. Geometric results derived from these data form the basis for the systematic acquisition of chemical and structural knowledge. This knowledge may be expressed in terms of numerical values (e.g. typical geometries), chemical and structural concepts (e.g. conformational descriptors), or as algorithmic relationships (e.g. correlations between parameters). A variety of knowledge acquisition methods that can help transform the crystallographic databases into knowledge bases will be illustrated. Conceptual Clustering Applications to Crystallographic Data : - ---------------------------------------------------- Suzanne Fortier ----------------- Our research in machine learning is motivated by the need for techniques to structure, manage and compress the rapidly growing crystallographic databases and transform them into knowledge bases. An incremental conceptual clustering algorithm, specifically designed for objects/scenes composed of many parts, has been designed and implemented. The algorithm and an initial application to pyranose sugar data will be described. Location: Facultes Universitaires Notre-Dame de la Paix - --------- Chemistry Department Auditorium CH2 Rue Grafe, 2 B-5000 NAMUR Belgium Schedule: November, the 19th, 1992 - --------- 15:00 Information: Prof. D. P. Vercauteren - ------------ Dr. L. Leherte E. Titeca Laboratoire de Physico-Chimie Informatique email:vercau,leherte,titeca at scf.fundp.ac.be Tel:+32-81-724534, +32-81-724535 Fax:+32-81-724530 Acknowledge-To: ------------------------------ Subject: New version of Self-Organizing Maps PD program package From: LVQ_PAK Date: Tue, 03 Nov 92 12:29:14 +0700 ************************************************************************ * * * SOM_PAK * * * * The * * * * Self-Organizing Map * * * * Program Package * * * * Version 1.2 (November 2, 1992) * * * * Prepared by the * * SOM Programming Team of the * * Helsinki University of Technology * * Laboratory of Computer and Information Science * * Rakentajanaukio 2 C, SF-02150 Espoo * * FINLAND * * * * Copyright (c) 1992 * * * ************************************************************************ Some time ago we released the software package "LVQ_PAK" for the easy application of Learning Vector Quantization algorithms. Corresponding public-domain programs for the Self-Organizing Map (SOM) algorithms are now available via anonymous FTP on the Internet. "What does the Self-Organizing Map mean?", you may ask --- See the following reference, then: Teuvo Kohonen. The self-organizing map. Proceedings of the IEEE, 78(9):1464-1480, 1990. In short, Self-Organizing Map (SOM) defines a 'non-linear projection' of the probability density function of the high-dimensional input data onto the two-dimensional display. SOM places a number of reference vectors into an input data space to approximate to its data set in an ordered fashion. This package contains all the programs necessary for the application of Self-Organizing Map algorithms in an arbitrary complex data visualization task. This code is distributed without charge on an "as is" basis. There is no warranty of any kind by the authors or by Helsinki University of Technology. In the implementation of the SOM programs we have tried to use as simple code as possible. Therefore the programs are supposed to compile in various machines without any specific modifications made on the code. All programs have been written in ANSI C. The programs are available in two archive formats, one for the UNIX-environment, the other for MS-DOS. Both archives contain exactly the same files. These files can be accessed via FTP as follows: 1. Create an FTP connection from wherever you are to machine "cochlea.hut.fi". The internet address of this machine is 130.233.168.48, for those who need it. 2. Log in as user "anonymous" with your own e-mail address as password. 3. Change remote directory to "/pub/som_pak". 4. At this point FTP should be able to get a listing of files in this directory with DIR and fetch the ones you want with GET. (The exact FTP commands you use depend on your local FTP program.) Remember to use the binary transfer mode for compressed files. The som_pak program package includes the following files: - Documentation: README short description of the package and installation instructions som_doc.ps documentation in (c) PostScript format som_doc.ps.Z same as above but compressed som_doc.txt documentation in ASCII format - Source file archives (which contain the documentation, too): som_p1r2.exe Self-extracting MS-DOS archive file som_pak-1.2.tar UNIX tape archive file som_pak-1.2.tar.Z same as above but compressed An example of FTP access is given below unix> ftp cochlea.hut.fi (or 130.233.168.48) Name: anonymous Password: ftp> cd /pub/som_pak ftp> binary ftp> get som_pak-1.2.tar.Z ftp> quit unix> uncompress som_pak-1.2.tar.Z unix> tar xvfo som_pak-1.2.tar See file README for further installation instructions. All comments concerning this package should be addressed to som@cochlea.hut.fi. ------------------------------ Subject: Audio Synthesizer From: warthman@garnet.berkeley.edu Date: Thu, 05 Nov 92 17:54:17 -0800 [[ Editor's Note: Ah, neural nets in "mainstream" art. Could someone give a few more (technical) details about the project? -PM ]] ********************** News Release ************************ November 5, 1992 ************************************************************ Neural-Network Audio Synthesizer Debuts at Paris Opera House ************************************************************ Palo Alto, California -- The old Opera House in Paris, France, will feature five performances by the Merce Cunningham Dance Company, November 12 to 17, in which a new type of audio synthesizer based on an artificial neural network will be used to generate electronic music. The synthesizer's musical accompaniment was composed and will be performed by David Tudor and his dance company colleague, Takehisa Kosugi. The audio synthesizer is built around an integrated-circuit chip from Intel Corporation in Santa Clara, California. The chip, called the Intel 80170NX electrically trainable analog neural network (ETANN), simulates the function of nerve cells in a biological brain. A remarkable range of audio effects can be generated with the electronic synthesizer -- from unique space-age and science-fiction sounds to passages that sound very much like birds, heart beats, porpoises, engines, and acoustical, percussion or string musical instruments. Sounds are generated internally by the synthesizer. External inputs such as voice, music, or random sounds can optionally be used to enrich or control the internally generated sounds. In addition to generating outputs to multiple audio speakers, the synthesizer can simultaneously drive oscilloscopes or other visual devices. The neural network chip's software consists of numeric values representing interconnection strengths between inputs and outputs -- a configuration analogous to the excitatory or inhibitory strengths of synapse connections between biological nerve cells. The artificial neurons can be connected in loops, using the programmable interconnection strengths, or they can be connected outside the chip with cables and feedback circuits. Audio oscillations occur as a result of delay in the feedback paths and thermal noise in the neural network chip. The sounds are generally rich because of the complexity of the circuitry. The concept for the synthesizer evolved from a project begun in 1989 by Forrest Warthman and David Tudor. The synthesizer was designed and built by Warthman; Mark Thorson, a hardware designer and associate editor of Microprocessor Report; and Mark Holler, Intel's program manager for neural network products. John Cage visited the design group in Palo Alto a few months before his passing away at the age of 79 this year. His observations on the synthesizer's role in musical composition and dance performance contributed to its current design. A description of the synthesizer's architecture and circuitry will appear in the February 1993 issue of Dr. Dobb's Journal. ------------------------------ Subject: NIPS*92 and CME travel From: Steve Hanson Date: Thu, 05 Nov 92 10:05:59 -0500 NIPS*92 Goers: This is a annoucement will we try and send out to you in the next week, but the date is so tight that I am sending it on the Net first. Please repost and send to your NIPS colleagues. Thanks. Steve Hanson NIPS*92 General Chair CME Travel (big mountain picture in background) INVITATION TO ROCKIES: On behalf of the NIPS Conference Coordinators, CME and CME Travel would like to welcome you to the Vail Valley. Your organization has selected Colorado Mountain Express to assit with your travel needs while attending the NEURAL INFORMATION PROCESSING SYSTEMS WORKSHOP at the Radissson Resort in Vail Colorado, December 2-5, 1992. In an effort to provide the most economic and professional service, speical discounted airfare and ground transportation rates have been negotiated to fly you into Denver and transfer you on December 3 at 1:30pm from Marriott's City Center Hotel to the Radisson in Vail and return you back to Denver Stapleton Airport upon your requested departure. Colorado Mountain Express located in the VAil Valley, has been serving the Vail and Beaver Creek Resort since 1983. Your speical group code "NIPS" not only provides you access to SPECIAL AIRLINE FARES, negotiaed on your behalf but also makes available preferred gound transfer rates with Colorado Mountona Express or Hertz Car Rental. ***NIPS*** Special Group Code ******Preferred Airline Contracts****** ******Discounted Ground Transportation**** via Colorado Mountain Express or Hertz Car Rental 1-800-525-6363 RSVP by NOVEMBER 18, 1992 We look forward to coordinating your travel arrangements. Please contact a CME travel Consultant at ext 6100 no later than Nov. 18th to secure your travel plans. Sincerely, Colorado Mountain Express & CME Travel Stephen J. Hanson Learning Systems Department SIEMENS Research 755 College Rd. East Princeton, NJ 08540 ------------------------------ Subject: Re: Hotel reservation deadline for NIPS workshops From: Steve Hanson Date: Mon, 02 Nov 92 09:22:14 -0500 Note that in the last 6 months that Mariott MARK Resort has been purchased by the Radisson and is now Called Radisson Vail Resort (same place, same facilties). If you have gotten reservations at the Mariott MARK Resort during this time under the NIPS*92 group, they will be honored by the Radisson. Everyone else who has yet to reserve a room at Vail, should call the Radisson as Gerry suggests ASAP. Steve NIPS*92 General Chair Stephen J. Hanson Learning Systems Department SIEMENS Research 755 College Rd. East Princeton, NJ 08540 ------------------------------ End of Neuron Digest [Volume 10 Issue 17] ***************************************** Received: from BUACCA by BUACCA.BU.EDU (Mailer R2.08 PTF009) with BSMTP id 5921; Thu, 12 Nov 92 20:29:23 EST Received: from noc2.dccs.upenn.edu by BUACCA.BU.EDU (IBM VM SMTP R1.2.1) with TCP; Thu, 12 Nov 92 20:29:14 EST Received: from CATTELL.PSYCH.UPENN.EDU by noc2.dccs.upenn.edu id AA13983; Thu, 12 Nov 92 20:24:52 -0500 Return-Path: Received: from LOCALHOST by cattell.psych.upenn.edu id AA00864; Thu, 12 Nov 92 19:21:52 EST Posted-Date: Thu, 12 Nov 92 19:21:25 EST From: "Neuron-Digest Moderator" To: Neuron-Distribution:; Subject: Neuron Digest V10 #18 (discussion + jobs + software) Reply-To: "Neuron-Request" X-Errors-To: "Neuron-Request" Organization: University of Pennsylvania Date: Thu, 12 Nov 92 19:21:25 EST Message-Id: <863.721614085@cattell.psych.upenn.edu> Sender: marvit@cattell.psych.upenn.edu Neuron Digest Thursday, 12 Nov 1992 Volume 10 : Issue 18 Today's Topics: Neural Nets Based Systems Re: Neural Nets Based Systems Re: Neural Nets Based Systems Stock Market Stock Market Help on hybrid systems Postdoc Position in Lund Free Neural Network Simulation and Analysis SW (am6.0) Send submissions, questions, address maintenance, and requests for old issues to "neuron-request@cattell.psych.upenn.edu". The ftp archives are available from cattell.psych.upenn.edu (130.91.68.31). Back issues requested by mail will eventually be sent, but may take a while. ---------------------------------------------------------------------- Subject: Neural Nets Based Systems From: ehf@philabs.philips.com (Eberhard Fisch) Organization: Philips Laboratories, Briarcliff, NY 10510 Date: 09 Nov 92 15:09:59 +0000 [[ Editor's Note: Given the recent post about a "stock market game," I gleaned this message and the following two from a mailing list devoted to general investing. Just For Your Information, of course. -PM ]] I am interested in investigating the potential of neural networks for forcasting markets. Im looking for articles or books dealing with the application of neural nets for forcasting market moves or finding patterns that occur in markets. I am not looking for, nor do I expect, neural nets or any other model for market analysis to be a magic solution. I just want to see whether neural network based systems have merit. Any opinions from people who have had some experience in using some of the standard neural network software or have developed their own software are welcome. Eberhard Fisch Philips Labs, NY ------------------------------ Subject: Re: Neural Nets Based Systems From: venky@thelonious.bellcore.com (G A Venkatesh) Organization: Bellcore, Morristown NJ Date: 09 Nov 92 18:56:52 +0000 Re: article by ehf@philabs.philips.com (Eberhard Fisch) writes: I don't know where you could find more information about them but there are at least two mutual funds (Fidelity Disciplined Equity and Fidelity Stock Selector) that use a neural net program to help with their buy/sell decisions. They seem to be doing well but I don't know how much the programs help. venky ------------------------------ Subject: Re: Neural Nets Based Systems From: mincy@think.com (Jeffrey Mincy) Organization: Thinking Machines Corporation, Cambridge MA, USA Date: 09 Nov 92 19:13:21 +0000 Re: article by ehf@philabs.philips.com (Eberhard Fisch) writes: The WSJ did the following article a few weeks ago: "Fidelity's Bradford Lewis Takes Aim at Indexes With His 'neural Network' Computer Program, by Robert Mcgough ... This fund is an index killer, he says of his fidelity disciplined equity fund. Mr Lewis leaves the stock-picking to a "neural network," a computer program that tries to mimic the intricate structure of the human brain. ... Disciplined equity has beaten the S&P 500 by 2.3 to 5.6 percentage points in each of three years to 1991. It's doint the same in the ytd with 5.8% return ...Only three other stock funds tracked by lipper did better than the market in all four of these periods. ... Anyway, perhaps this gives you some information. -- jeff mincy@think.com ------------------------------ Subject: Stock Market From: D.Navin.Chandra@ISL1.RI.CMU.EDU Date: Sun, 08 Nov 92 20:26:40 -0500 [[ Editor's Note: See previous and following note about the use of Neural Networks in the stock market. It should be noted also that various "dart board" investment strategoes regularly outperform the profession money managers. See also Vol 10 #11 and #13 for related bibliographies. -PM ]] Gary Bradsky I agree the announcement was wrong, but please dont belittle the things that other people try to work on. The Stock Market game is a legitimate game entered into by 1000's of students from across the country. Many people use NNets and GA's to "crack" the market trend. The area has also been the most successful application of NN to date. navin ------------------------------ Subject: Stock Market From: bradski@cns.bu.edu Date: Mon, 09 Nov 92 21:01:49 -0500 [[ Re: previous message from D.Navin.Chandra@ISL1.RI.CMU.EDU ]] If the stock market game had not been a money making scheme, but simply an "educational" contest as advertised, I would not have complained at all. I also don't personally care if some people make this type of contest a money making business -- just let them advertise in magazines, or TV, not in a news group (I get enough ads as it is). Now, to add increasing amounts of content: (1) APPLICATIONS OF NN: Stock market price prediction is not "the most successful application of NN to date", the most successful application of neural networks is clearly in adaptive filtering -- particularly for echo cancellation. Adalines or their variants (lattice filters etc...) are used in nearly every long distance phone line, modems, in EKG, EEG, ultra sound and the list goes on. (2) PREDICTION, STOCKS, AND SELECTIVE INFORMATION: Neuron readers ought to be aware, that even if someone "cracks" the market with their NN application and wins a stock market contest, it means absolutely nothing more than if someone won the contest by throwing darts at the Wall St Journal and looking for the holes on the stock quotation pages. It means nothing more because we know nothing about the statistical sample out of which we hear about the "winner" (and never the losers). If 5000 people try to predict stocks with backprop and one of them wins big, and the rest fall around the mean of stock performance, should you put your *future* money with the one winner? Should you use backprop? Or, did the winner just happen to hit on a lucky fit of the now, past data, that tells you nothing about future performance? (3) PREDICTING MARKET TRENDS MAY BE IN NEED OF BELITTLEMENT: Trying to "crack the market trend" may in fact be hopeless, if by "trend" one means the future expected value of stock prices. If you are going to use NNs with financial data, get an intro into financial theory (but be prepared for a little math). I suggest: Easy intro: John Hull, "Options, Futures, and Other Derivative Securities", Prentice-Hall, 1989, More cutting edge, but still decipherable: "Continuous-Time Finance", R.C. Merton, Basil Blackwell, 1992 (get the cheaper paperback version). In Continuous-Time Finance, read chapter 3 for an easy development of stochastic calculus and Ito's lemma. In chapter 3, Merton uses "order statistics" to develop his equations (order statistics just tell you how fast a term blows up or goes to zero). On page 68-9 he develops a model of stock market price changes based on an Ito process: X(t) - X(t-1) = mean(t)*h + std_dev(t)*noise*h^(1/2) (A) where h is a small time increment which goes to zero in the limit, and noise can be taken as zero-mean Gaussian. The point of all this is that in a small increment of time, the h^(1/2) "noise" or variance term dominates the mean. This is observed in practice, eg. the variability of a stock's price swamps out the level, or expected value of its price. Moreover to estimate the mean of (A) over n observations over a total time period T (T = h*n), we'd use: est_mean = Sum_{k=1 to n} (X(k) - X(k-1))/T = mean(t) In other words, the estimate of the mean is not affected by choosing finer observation intervals, only in the length of time "T" that we measure over. For estimating the variance of (A): est_var = (Sum_{k=1 to n} (X(k) - X(k-1))^2/T = std_dev^2 + h*mean^2 = std_dev^2 + (T/n)*mean^2 becomes better and better with finer measurements (larger n). THE CONCLUSION: variances (or covariances) are easier to estimate from stochastic time series than are the means. This is a fundamental fact, which no algorithm, no matter how neural or not, can overcome. WHAT IT "MEANS": The stock and option models developed by Black-Scholes, Merton and others have shown themselves to be useful (so useful in fact that they are essentially what are keeping our large banks profitable now). These models don't depend on having to know the mean, but do depend on having to estimate the variance of stocks. THUS, if you want to come up with a useful NN financial application, try using NNs to estimate stock variance, not stock prices. You might start by using adaptive filters as per (1) above to do this. --Gary Bradski (bradski@cns.bu.edu) ------------------------------ Subject: Help on hybrid systems From: Fernando Passold Organization: Universidade Federal de Santa Catarina/BRASIL Date: Thu, 12 Nov 92 17:37:23 -0300 [[ Editor's Note: I assume this person wanted his message to be published. If you, loyal reader, feel *your* reply to this fellow might be of more general interest, please send a copy to neuron-request@cattell.psych.upenn.edu -PM ]] I would like to know the intend of this list and if I could obtain some replies. I am a researcher tacking part of Biomedical Engineering Group Laboratory of Electrical Engineering Department at the Federal University of Santa Catarina. We develop little biomedical equipments and Expest Systems - such a Hospital Infection Control Aid and a Hybrid System (rule-based and neural network based) for Proposal and Evaluation of Anesthesian Plan. I am a master degree student, now, grapple in develop a Expert Network (also called, hybrid system: those whom combine neural networks with rule-based methods) for Planning and Evaluation of Plans of Anesthesia, continuing a PHD thesis, for Critical Patients (their who need critical cares) or Problem Patients (exceptional cases of patientes whom evaluate to critical patients). I have some troubles choosing the most suitable approach to develop this system. I do not know if neural networks could be the key to solve the main part of the problem, because we are dealing with exceptions that probably could best solved throught a rule-based method. At lately, there are a few shells systems applying object-orienthed technics with rule-based or frames methods such the new Kappa PC Application Development Systems for Windows environment, from IntellCorp. Inc.. So, I am interested in implementations of hybrid systems using object-oriented programming. Maybe, there will be a way to link neural networks simulators using object-oriented approach to another heuristical languages such as Turbo-Prolog object- oriented. I would be glad if someone could make a comment about it, as well as indicate if there exists a _similar research_. Please, reply directly to me as I am not subcribed to this list. Thanks a lot respect to this matter, best regards, Fernando Passold Biomedical Engineering Group Lab. UFSC/BRAZIL E-mail: ee3fps@brufsc.bitnet P.S.: I am interested in subcribing in your list if not a large amount of material is normaly posted to it. ------------------------------ Subject: Postdoc Position in Lund From: carsten@thep.lu.se Date: Tue, 10 Nov 92 15:01:40 +0100 A two year postdoc position will be available within the Complex Systems group at the Department of Theoretical Physics, University of Lund, Sweden, starting September 1st 1993. The major research area of the group is Artificial Neural Networks with tails into chaos and difficult computational problems in general. Although some application studies occur, algorithmic development is the focus in particular within the following areas: * Using Feed-back ANN for finding good solutions to combinatorial optimization problems; knapsacks, scheduling, track-finding. * Time-series prediction. * Robust multi-layer perceptron updating procedures including noise. * Deformable template methods -- robust statistics. * Configurational Chemistry -- Polymers, Proteins ... * Application work within the domain of experimental physics, in particular in connection with the upcoming SSC/LHC experiments. Lund University is the largest campus in Scandinavia located in a picturesque 1000 year old city (100k inhabitants). Lund is strategically well located in the south of Sweden with 1.5 hrs commuting distance to Copenhagen (Denmark). The candidate should have a PhD in a relevant field, which need not be Physics/Theoretical Physics. Applications and three letters of recommendation should be sent to (not later than December 15): Carsten Peterson Department of Theoretical Physics University of Lund Solvegatan 14A S-223 62 Lund Sweden or Bo S\"{o}derberg Department of Theoretical Physics University of Lund Solvegatan 14A S-223 62 Lund Sweden ------------------------------ Subject: Free Neural Network Simulation and Analysis SW (am6.0) From: Russell R Leighton Date: Fri, 30 Oct 92 09:09:54 -0500 ************************************************************************* **** delete all prerelease versions!!!!!!! (they are not up to date) **** ************************************************************************* The following describes a neural network simulation environment made available free from the MITRE Corporation. The software contains a neural network simulation code generator which generates high performance ANSI C code implementations for modular backpropagation neural networks. Also included is an interface to visualization tools. FREE NEURAL NETWORK SIMULATOR AVAILABLE Aspirin/MIGRAINES Version 6.0 The Mitre Corporation is making available free to the public a neural network simulation environment called Aspirin/MIGRAINES. The software consists of a code generator that builds neural network simulations by reading a network description (written in a language called "Aspirin") and generates an ANSI C simulation. An interface (called "MIGRAINES") is provided to export data from the neural network to visualization tools. The previous version (Version 5.0) has over 600 registered installation sites world wide. The system has been ported to a number of platforms: Host platforms: convex_c2 /* Convex C2 */ convex_c3 /* Convex C3 */ cray_xmp /* Cray XMP */ cray_ymp /* Cray YMP */ cray_c90 /* Cray C90 */ dga_88k /* Data General Aviion w/88XXX */ ds_r3k /* Dec Station w/r3000 */ ds_alpha /* Dec Station w/alpha */ hp_parisc /* HP w/parisc */ pc_iX86_sysvr4 /* IBM pc 386/486 Unix SysVR4 */ pc_iX86_sysvr3 /* IBM pc 386/486 Interactive Unix SysVR3 */ ibm_rs6k /* IBM w/rs6000 */ news_68k /* News w/68XXX */ news_r3k /* News w/r3000 */ next_68k /* NeXT w/68XXX */ sgi_r3k /* Silicon Graphics w/r3000 */ sgi_r4k /* Silicon Graphics w/r4000 */ sun_sparc /* Sun w/sparc */ sun_68k /* Sun w/68XXX */ Coprocessors: mc_i860 /* Mercury w/i860 */ meiko_i860 /* Meiko w/i860 Computing Surface */ Included with the software are "config" files for these platforms. Porting to other platforms may be done by choosing the "closest" platform currently supported and adapting the config files. New Features - ------------ - ANSI C ( ANSI C compiler required! If you do not have an ANSI C compiler, a free (and very good) compiler called gcc is available by anonymous ftp from prep.ai.mit.edu (18.71.0.38). ) Gcc is what was used to develop am6 on Suns. - Autoregressive backprop has better stability constraints (see examples: ringing and sequence), very good for sequence recognition - File reader supports "caching" so you can use HUGE data files (larger than physical/virtual memory). - The "analyze" utility which aids the analysis of hidden unit behavior (see examples: sonar and characters) - More examples - More portable system configuration for easy installation on systems without a "config" file in distribution Aspirin 6.0 - ------------ The software that we are releasing now is for creating, and evaluating, feed-forward networks such as those used with the backpropagation learning algorithm. The software is aimed both at the expert programmer/neural network researcher who may wish to tailor significant portions of the system to his/her precise needs, as well as at casual users who will wish to use the system with an absolute minimum of effort. Aspirin was originally conceived as ``a way of dealing with MIGRAINES.'' Our goal was to create an underlying system that would exist behind the graphics and provide the network modeling facilities. The system had to be flexible enough to allow research, that is, make it easy for a user to make frequent, possibly substantial, changes to network designs and learning algorithms. At the same time it had to be efficient enough to allow large ``real-world'' neural network systems to be developed. Aspirin uses a front-end parser and code generators to realize this goal. A high level declarative language has been developed to describe a network. This language was designed to make commonly used network constructs simple to describe, but to allow any network to be described. The Aspirin file defines the type of network, the size and topology of the network, and descriptions of the network's input and output. This file may also include information such as initial values of weights, names of user defined functions. The Aspirin language is based around the concept of a "black box". A black box is a module that (optionally) receives input and (necessarily) produces output. Black boxes are autonomous units that are used to construct neural network systems. Black boxes may be connected arbitrarily to create large possibly heterogeneous network systems. As a simple example, pre or post-processing stages of a neural network can be considered black boxes that do not learn. The output of the Aspirin parser is sent to the appropriate code generator that implements the desired neural network paradigm. The goal of Aspirin is to provide a common extendible front-end language and parser for different network paradigms. The publicly available software will include a backpropagation code generator that supports several variations of the backpropagation learning algorithm. For backpropagation networks and their variations, Aspirin supports a wide variety of capabilities: 1. feed-forward layered networks with arbitrary connections 2. ``skip level'' connections 3. one and two-dimensional weight tessellations 4. a few node transfer functions (as well as user defined) 5. connections to layers/inputs at arbitrary delays, also "Waibel style" time-delay neural networks 6. autoregressive nodes. 7. line search and conjugate gradient optimization The file describing a network is processed by the Aspirin parser and files containing C functions to implement that network are generated. This code can then be linked with an application which uses these routines to control the network. Optionally, a complete simulation may be automatically generated which is integrated with the MIGRAINES interface and can read data in a variety of file formats. Currently supported file formats are: Ascii Type1, Type2, Type3 Type4 Type5 (simple floating point file formats) ProMatlab Examples - -------- A set of examples comes with the distribution: xor: from RumelHart and McClelland, et al, "Parallel Distributed Processing, Vol 1: Foundations", MIT Press, 1986, pp. 330-334. encode: from RumelHart and McClelland, et al, "Parallel Distributed Processing, Vol 1: Foundations", MIT Press, 1986, pp. 335-339. bayes: Approximating the optimal bayes decision surface for a gauss-gauss problem. detect: Detecting a sine wave in noise. iris: The classic iris database. characters: Learing to recognize 4 characters independent of rotation. ring: Autoregressive network learns a decaying sinusoid impulse response. sequence: Autoregressive network learns to recognize a short sequence of orthonormal vectors. sonar: from Gorman, R. P., and Sejnowski, T. J. (1988). "Analysis of Hidden Units in a Layered Network Trained to Classify Sonar Targets" in Neural Networks, Vol. 1, pp. 75-89. spiral: from Kevin J. Lang and Michael J, Witbrock, "Learning to Tell Two Spirals Apart", in Proceedings of the 1988 Connectionist Models Summer School, Morgan Kaufmann, 1988. ntalk: from Sejnowski, T.J., and Rosenberg, C.R. (1987). "Parallel networks that learn to pronounce English text" in Complex Systems, 1, 145-168. perf: a large network used only for performance testing. monk: The backprop part of the monk paper. The MONK's problem were the basis of a first international comparison of learning algorithms. The result of this comparison is summarized in "The MONK's Problems - A Performance Comparison of Different Learning algorithms" by S.B. Thrun, J. Bala, E. Bloedorn, I. Bratko, B. Cestnik, J. Cheng, K. De Jong, S. Dzeroski, S.E. Fahlman, D. Fisher, R. Hamann, K. Kaufman, S. Keller, I. Kononenko, J. Kreuziger, R.S. Michalski, T. Mitchell, P. Pachowicz, Y. Reich H. Vafaie, W. Van de Welde, W. Wenzel, J. Wnek, and J. Zhang has been published as Technical Report CS-CMU-91-197, Carnegie Mellon University in Dec. 1991. wine: From the ``UCI Repository Of Machine Learning Databases and Domain Theories'' (ics.uci.edu: pub/machine-learning-databases). Performance of Aspirin simulations - ---------------------------------- The backpropagation code generator produces simulations that run very efficiently. Aspirin simulations do best on vector machines when the networks are large, as exemplified by the Cray's performance. All simulations were done using the Unix "time" function and include all simulation overhead. The connections per second rating was calculated by multiplying the number of iterations by the total number of connections in the network and dividing by the "user" time provided by the Unix time function. Two tests were performed. In the first, the network was simply run "forward" 100,000 times and timed. In the second, the network was timed in learning mode and run until convergence. Under both tests the "user" time included the time to read in the data and initialize the network. Sonar: This network is a two layer fully connected network with 60 inputs: 2-34-60. Millions of Connections per Second Forward: SparcStation1: 1 IBM RS/6000 320: 2.8 HP9000/720: 4.0 Meiko i860 (40MHz) : 4.4 Mercury i860 (40MHz) : 5.6 Cray YMP: 21.9 Cray C90: 33.2 Forward/Backward: SparcStation1: 0.3 IBM RS/6000 320: 0.8 Meiko i860 (40MHz) : 0.9 HP9000/720: 1.1 Mercury i860 (40MHz) : 1.3 Cray YMP: 7.6 Cray C90: 13.5 Gorman, R. P., and Sejnowski, T. J. (1988). "Analysis of Hidden Units in a Layered Network Trained to Classify Sonar Targets" in Neural Networks, Vol. 1, pp. 75-89. Nettalk: This network is a two layer fully connected network with [29 x 7] inputs: 26-[15 x 8]-[29 x 7] Millions of Connections per Second Forward: SparcStation1: 1 IBM RS/6000 320: 3.5 HP9000/720: 4.5 Mercury i860 (40MHz) : 12.4 Meiko i860 (40MHz) : 12.6 Cray YMP: 113.5 Cray C90: 220.3 Forward/Backward: SparcStation1: 0.4 IBM RS/6000 320: 1.3 HP9000/720: 1.7 Meiko i860 (40MHz) : 2.5 Mercury i860 (40MHz) : 3.7 Cray YMP: 40 Cray C90: 65.6 Sejnowski, T.J., and Rosenberg, C.R. (1987). "Parallel networks that learn to pronounce English text" in Complex Systems, 1, 145-168. Perf: This network was only run on a few systems. It is very large with very long vectors. The performance on this network is in some sense a peak performance for a machine. This network is a two layer fully connected network with 2000 inputs: 100-500-2000 Millions of Connections per Second Forward: Cray YMP 103.00 Cray C90 220 Forward/Backward: Cray YMP 25.46 Cray C90 59.3 MIGRAINES - ------------ The MIGRAINES interface is a terminal based interface that allows you to open Unix pipes to data in the neural network. This replaces the NeWS1.1 graphical interface in version 4.0 of the Aspirin/MIGRAINES software. The new interface is not a simple to use as the version 4.0 interface but is much more portable and flexible. The MIGRAINES interface allows users to output neural network weight and node vectors to disk or to other Unix processes. Users can display the data using either public or commercial graphics/analysis tools. Example filters are included that convert data exported through MIGRAINES to formats readable by: - Gnuplot 3 - Matlab - Mathematica - Xgobi Most of the examples (see above) use the MIGRAINES interface to dump data to disk and display it using a public software package called Gnuplot3. Gnuplot3 can be obtained via anonymous ftp from: >>>> In general, Gnuplot 3 is available as the file gnuplot3.?.tar.Z >>>> Please obtain gnuplot from the site nearest you. Many of the major ftp >>>> archives world-wide have already picked up the latest version, so if >>>> you found the old version elsewhere, you might check there. >>>> >>>> NORTH AMERICA: >>>> >>>> Anonymous ftp to dartmouth.edu (129.170.16.4) >>>> Fetch >>>> pub/gnuplot/gnuplot3.?.tar.Z >>>> in binary mode. >>>>>>>> A special hack for NeXTStep may be found on 'sonata.cc.purdue.edu' >>>>>>>> in the directory /pub/next/submissions. The gnuplot3.0 distribution >>>>>>>> is also there (in that directory). >>>>>>>> >>>>>>>> There is a problem to be aware of--you will need to recompile. >>>>>>>> gnuplot has a minor bug, so you will need to compile the command.c >>>>>>>> file separately with the HELPFILE defined as the entire path name >>>>>>>> (including the help file name.) If you don't, the Makefile will over >>>>>>>> ride the def and help won't work (in fact it will bomb the program.) NetTools - ----------- We have include a simple set of analysis tools by Simon Dennis and Steven Phillips. They are used in some of the examples to illustrate the use of the MIGRAINES interface with analysis tools. The package contains three tools for network analysis: gea - Group Error Analysis pca - Principal Components Analysis cda - Canonical Discriminants Analysis Analyze - ------- "analyze" is a program inspired by Denis and Phillips' Nettools. The "analyze" program does PCA, CDA, projections, and histograms. It can read the same data file formats as are supported by "bpmake" simulations and output data in a variety of formats. Associated with this utility are shell scripts that implement data reduction and feature extraction. "analyze" can be used to understand how the hidden layers separate the data in order to optimize the network architecture. How to get Aspirin/MIGRAINES - ----------------------- The software is available from two FTP sites, CMU's simulator collection and UCLA's cognitive science machines. The compressed tar file is a little less than 2 megabytes. Most of this space is taken up by the documentation and examples. The software is currently only available via anonymous FTP. > To get the software from CMU's simulator collection: 1. Create an FTP connection from wherever you are to machine "pt.cs.cmu.edu" (128.2.254.155). 2. Log in as user "anonymous" with password your username. 3. Change remote directory to "/afs/cs/project/connect/code". Any subdirectories of this one should also be accessible. Parent directories should not be. ****You must do this in a single operation****: cd /afs/cs/project/connect/code 4. At this point FTP should be able to get a listing of files in this directory and fetch the ones you want. Problems? - contact us at "connectionists-request@cs.cmu.edu". 5. Set binary mode by typing the command "binary" ** THIS IS IMPORTANT ** 6. Get the file "am6.tar.Z" > To get the software from UCLA's cognitive science machines: 1. Create an FTP connection to "ftp.cognet.ucla.edu" (128.97.50.19) (typically with the command "ftp ftp.cognet.ucla.edu") 2. Log in as user "anonymous" with password your username. 3. Change remote directory to "alexis", by typing the command "cd alexis" 4. Set binary mode by typing the command "binary" ** THIS IS IMPORTANT ** 5. Get the file by typing the command "get am6.tar.Z" Other sites - ----------- If these sites do not work well for you, then try the archie internet mail server. Send email: To: archie@cs.mcgill.ca Subject: prog am6.tar.Z Archie will reply with a list of internet ftp sites that you can get the software from. How to unpack the software - -------------------------- After ftp'ing the file make the directory you wish to install the software. Go to that directory and type: zcat am6.tar.Z | tar xvf - -or- uncompress am6.tar.Z ; tar xvf am6.tar How to print the manual - ----------------------- The user documentation is located in ./doc in a few compressed PostScript files. To print each file on a PostScript printer type: uncompress *.Z lpr -s *.ps Why? - ---- I have been asked why MITRE is giving away this software. MITRE is a non-profit organization funded by the U.S. federal government. MITRE does research and development into various technical areas. Our research into neural network algorithms and applications has resulted in this software. Since MITRE is a publically funded organization, it seems appropriate that the product of the neural network research be turned back into the technical community at large. Thanks - ------ Thanks to the beta sites for helping me get the bugs out and make this portable. Thanks to the folks at CMU and UCLA for the ftp sites. Copyright and license agreement - ------------------------------- Since the Aspirin/MIGRAINES system is licensed free of charge, the MITRE Corporation provides absolutely no warranty. Should the Aspirin/MIGRAINES system prove defective, you must assume the cost of all necessary servicing, repair or correction. In no way will the MITRE Corporation be liable to you for damages, including any lost profits, lost monies, or other special, incidental or consequential damages arising out of the use or in ability to use the Aspirin/MIGRAINES system. This software is the copyright of The MITRE Corporation. It may be freely used and modified for research and development purposes. We require a brief acknowledgement in any research paper or other publication where this software has made a significant contribution. If you wish to use it for commercial gain you must contact The MITRE Corporation for conditions of use. The MITRE Corporation provides absolutely NO WARRANTY for this software. October, 1992 Russell Leighton * * MITRE Signal Processing Center *** *** *** *** 7525 Colshire Dr. ****** *** *** ****** McLean, Va. 22102, USA ***************************************** ***** *** *** ****** INTERNET: taylor@world.std.com, ** *** *** *** leighton@mitre.org * * ------------------------------ End of Neuron Digest [Volume 10 Issue 18] ***************************************** Received: from BUACCA by BUACCA.BU.EDU (Mailer R2.08 PTF009) with BSMTP id 7800; Thu, 19 Nov 92 13:33:38 EST Received: from noc2.dccs.upenn.edu by BUACCA.BU.EDU (IBM VM SMTP R1.2.1) with TCP; Thu, 19 Nov 92 13:33:35 EST Received: from CATTELL.PSYCH.UPENN.EDU by noc2.dccs.upenn.edu id AA17704; Thu, 19 Nov 92 12:28:13 -0500 Return-Path: Received: from LOCALHOST by cattell.psych.upenn.edu id AA25299; Thu, 19 Nov 92 11:18:37 EST Posted-Date: Thu, 19 Nov 92 11:18:06 EST From: "Neuron-Digest Moderator" To: Neuron-Distribution:; Subject: Neuron Digest V10 #19 (discussion + jobs) Reply-To: "Neuron-Request" X-Errors-To: "Neuron-Request" Organization: University of Pennsylvania Date: Thu, 19 Nov 92 11:18:06 EST Message-Id: <25295.722189886@cattell.psych.upenn.edu> Sender: marvit@cattell.psych.upenn.edu Neuron Digest Thursday, 19 Nov 1992 Volume 10 : Issue 19 Today's Topics: Re: Help on hybrid systems Stock Market Faculty position at NYU 4 research posts Research enquire FYI: Results of Physics of Computation Workshop Send submissions, questions, address maintenance, and requests for old issues to "neuron-request@cattell.psych.upenn.edu". The ftp archives are available from cattell.psych.upenn.edu (130.91.68.31). Back issues requested by mail will eventually be sent, but may take a while. ---------------------------------------------------------------------- Subject: Re: Help on hybrid systems From: Dave Cornforth Date: Fri, 13 Nov 92 10:55:40 +0000 Don't know much about your field, but here's my two cents worth... If you already have some firm rules which are sound, then why not use them directly in your system? If you have some decisions which are a bit more "wooley", then you could look at a neural network to cope with those. However, I would suggest you do some kind of statistical analysis on your data first. If you find that you have data which is roughly normally distributed and unimodal for each class, which you want to assign to different classes (or decisions), then you can't beat a Gaussian model Bayesian classifier. If the data is not normally distributed, or multi-modal, then you could look at "real" neural networks, but there are also some jolly good memory table-based classifiers which tend to be a lot quicker to train. Dave ------------------------------ Subject: Stock Market From: P.Refenes@cs.ucl.ac.uk Date: Fri, 13 Nov 92 10:04:10 +0000 I am afraid I have not followed this topic very closely, but here are some references of our own work in teh field: [1] Refenes A. N., Zapranis A., and Azema-Barac M., "Stock Ranking: Neural networks Versus Multiple linear Regression", Proc ICNN'93 San Francisco (submitted 1992). [2] Refenes A. N., et al "Currency Exchange rate prediction and Neural Network Design Strategies", Neural computing & Applications Journal, Vol 1, no. 1., (1993). [3] Refenes A. N., & Zaidi A., "Managing Exchange Rate Prediction Strategies with Neural Networks", Proc. Workshop on Neural Networks: techniques & Applications, Liverpool (Sept. 1992), also in Lisboa P. G., and Taylor M, "Neural Networks: techniques & Applications", Ellis Horwood (1992). [4] Refenes A. N., & Azema-Barac M., "Neural Networks for Tactical Asset Allocation in the Global Bonds Markets", Proc. IEE Third International Conference on ANNS, Brighton 1993 (submitted 1992). [5] Refenes A. N. et al "Financial Modelling using Neural Networks", in Liddell H. (ed) "Commercial Parallel Processing", Unicom, (to appear). ------------------------------ Subject: Faculty position at NYU From: ltm@cns.nyu.edu (Laurence T. Maloney) Date: Sun, 15 Nov 92 20:02:07 -0500 N N Y Y U U * NN N Y Y U U * * * N N N Y U U * * * N NN Y U U * N N Y U U U JOB OPPORTUNITY TENURED OR TENURE-TRACK COGNITIVE SCIENCE/COGNITIVE PSYCHOLOGY/COGNITIVE NEUROSCIENCE -- NEW YORK UNIVERSITY: The Department of Psychology, Faculty of Arts and Science, New York University anticipates making faculty appointments beginning September 1993, in one or more of the above areas. These are tenured or tenure-track position and the rank is open. Candidates must be engaged in an active research program at the forefront of modern cognitive psychology, cognitive science, or cognitive neuroscience. Senior applicants should submit a vita. Others should also include reprints of recent publications and three letters of reference. Applications should be sent to Professor Murray Glanzer, Department of Psychology, 6 Washington Place, Room 965. TO ENSURE CONSIDERATION, APPLICATIONS SHOULD BE RECEIVED BY JANUARY 15, 1993. New York University is an Equal Opportunity/Affirmative Action Employer. Brief questions? E-mail to Larry Maloney, ltm@cns.nyu.edu. ------------------------------ Subject: 4 research posts From: Noel Sharkey Date: Mon, 16 Nov 92 11:36:21 +0000 Research Posts at Centre for Connection Science Department of Computer Science University of Exeter UK Four new research posts will be available (expected start January 1st, 1993) at the Centre for Connection Science, Department of Computer Science, as part of a 3-year research project funded by the SERC/DTI and led by Noel Sharkey and Derek Partridge. The project will investigate the reliability of software systems implemented as neural nets using the multiversion programming strategy. Two of the posts will at the post-doctoral level (grade 1A). The ideal applicant will be proficient in both neural computing and software engineering (although training in one or the other may be given). In addition, there is a requirement for at least one of the successful applicants to work on the formal analysis of network implementation as a paradigm for reliable software. The other two posts will be for Research Assistants/Experimental Officers at grade 1b. One of these will be required to have a high level of proficiency in C programming and general computing skills. The other will be part-time, and preference will be given to an applicant with good mathematical and engineering skills (particulary control systems). For more information please contact Lyn Shackelton by email (lyn@dcs.ex.ac.uk or by telephone (0392-264066 mornings 10.00-2.00). ------------------------------ Subject: Research enquire From: LHOTAK@whmain.east-london.ac.uk Date: 18 Nov 92 18:29:57 +0000 Dear reader, My name is Martin Lhotak and I currenly became a PhD. student at the University of East London. The topic of my PhD concerns with a model of an artificial intellect based upon classification paradigm and self- organizing hierarchy establishment and a short proposal summary of my PhD. topic follows: =< Start of my PhD. proposal>=================================== Title: A model of an artificial intellect based upon classification paradigm and self-organizing hierarchy establishment The PhD. research would investigate the principles of human intellect and the generic intellectual processes in systems leading to qualified understanding of the world the system is based in. Current achievements in this area have recorded just a partial success due to the approach to modell the intellectual process by large amount of rules, data, formulas,etc., which tends to be valid only under specific conditions. In my understanding those attributes (facts, data, formulas) do not represent intelligence, but only support and monitor the development process of a given intellect. In my PhD research I would like to look at the aforementioned process from the point of view of how the organization of a system changes during the information gathering process and how significant it is to gather information from various independent (often contradictory) sources. Especially, I would like to focus on the essential principles, e.g. classification and hierarchization, which, I believe, helps human intellect to cope with complex tasks. Based upon the theory of object classification and the results observed from the general information gathering process I would like to compose a prototype of an adaptive intelligent system capable of autonomous classification and autonomous establishment of hierarchies in order to "understand" the meanings of given "model world" objects. The theoretical model would require continuous testing on a given set of subject domains, e.g. Plant Fossil Record - a large heterogeneous data set, etc. ==< End of my proposal >======================================== As you have quite likely observed my PhD. topic is rather generous (and perhaps too ambitious) and it is highly probable that I would be more specific on the PhD. registration form after I would do some preliminary research. I realize that the aforementioned proposal is fairly vague and "nothing-specific-saying", however, some essential ideas should be visible and I would be very grateful for any (especially critical) comments on my approach or proposal for my PhD. studies and any information whether anyone has tried (or rejected) similar approach or whether anyone is already doing similar research in this area. I would be also very thankful for any possibility of consulting and discussing my ideas directly, via e-mail, or in a sort of a larger scientific arena (listservers?, meetings?). The best way how to contact me is via e-mail: lhotak@whmain.uel.ac.uk Best Regards, o #/- Martin Lhotak # \-/| #/-\ | ==============================================================# \-/ Martin Lhotak # eMail: lhotak@whmain.uel.ac.uk # Systems & Computing # sMail: 39, Lodge Avenue # University of East London # Dagenham, RM8 2JD # United Kingdom # telNo: 081-590-7722 ext. 4101 # =============================================================== ------------------------------ Subject: FYI: Results of Physics of Computation Workshop From: Doug Matzke Date: Thu, 12 Nov 92 19:12:54 -0600 Hello The Physics of Computation Workshop was held on October 2-4, 1992 in Dallas Texas, and was attended by almost 100 people from industry, press, university, and government agencies. The attendees decided to: 1) Publish a post-proceedings of the papers with IEEE Press (see order form below to reserve you own copy - about 500 pages) 2) Start an electronic mailing list on the physics of computation. (Please send requests to physics.computation-request@hc.ti.com) 3) Have the next conference in two years. 4) Start publishing general articles to establish the field. One of the attendees is a science writer for the Dallas Morning News, and the text of his article is included to give you some sense of what happened at the workshop. So get involved with this emerging field by signing up with the mailing list, ordering your own copy of the post-proceedings, and making plans to attend the next conference. Doug Matzke Workshop Chairman EMAIL: matzke@hc.ti.com PHONE: (214) 995-0787 ********************************************************************* Order information for Post-Proceedings for Physics of Computation Workshop (Held on October 2-4, 1992 Dallas Texas) These post-proceedings are available at the prepublication rate of $35 (US Dollars), if your order is received by December 14, 1992, along with this form. If you order at the the conference rate, the book will be mailed to you directly from the printer. The Post-Proceedings will be available for purchase after publication, from the IEEE Computer Society Press, probably at a higher cost. NAME: ________________________________________________________________ ADDRESS:______________________________________________________________ CITY:___________________________________________ STATE:_______________ COUNTRY: _____________________________________ ZIPCODE:_______________ COPIES: ______________ X $35.00 = TOTAL AMOUNT: ______________________ Include check or money order for the full amount, made out to: Dallas IEEE Computer Society (No credit cards or COD, please.) Send this Completed form with check or money order to: Douglas Matzke Post-Proceedings Order for PCW Texas Instruments P.O. BOX 655474, MS 446 Dallas, Tx 75265 Questions may be directed to Doug at (214)995-0787 or matzke@hc.ti.com ********************************************************************* As appeared in Dallas Morning News Monday October 26, 1992 OPENING THE DOOR TO COMPUTER PHYSICS New field may solve mysteries about universe, time, genes By Tom Siegfried Some physicists can't get their minds off computers. Perhaps that's because minds seem to work a lot like like computers. Understanding the physics of how computers compute, some scientists reason, might provide clues to how brains do it, too. And even if it can't explain the the brain, the physics of computing may solve mysteries about the universe, the genetic code and why time always flows in one direction. At least those were among the topics discussed this month when researchers from around the world met in Addison to explore links between basic physical laws and the process of computation. Nearly 100 researchers from physics, computer science, mathematics and related fields shared insights on computers, quantum physics, black holes and the brain. Some speakers discussed using information theory to study problems in molecular biology or the working of human memory. Others showed how quantum theory can be used to devise better secret codes. Some speakers even described real physics in real computers. After all, Texas Instruments Inc. of Dallas sponsored the conference. At one level, studying the physics of computing does have a practical aspect. Knowing basic physics is critical in designing smaller, faster computer chips that won't melt down by producing too much heat. But speakers at the Addison meeting often focused on more theoretical questions, ranging from how to design a foolproof code for automates teller bank cards to whether the universe is a gigantic computer simulation. The diversity of the discussions suggested that while a new scientific discipline has been born, it's too soon for a christening. The new field is related to information theory, computer science, artificial intelligence, chaos and fractals, but it doesn't really have a name of its own. "It's clear that this field is not well defined," said Tommaso Toffoli of the computer science labaratory at the Massachusetts Institute of Technology. Researchers addressing this identity crisis could only conclude that the new field concerns itself with "fundamental connections between physics and computation." At the heart of this connection is an obscure but important rule called the Landauer principle. It involves how much energy a computation requires. More than three decades ago, IBM computer physicist Rolf Landauer calculated the least amount of energy needed to perform a single computational step. The answer, he showed, is essentially zero. If a computation is performed slowly eneough, the amount of energy used up can be as small as desired. That was a surprising result. Transferring information in a computer is like sending a message from one memory location to another, and physicists used to think that sending messages required energy. But the early studies took too limited a view of information transfer, usually considering signals sent by waves. "We don't have to send information by waves," said Dr. Landauer. "I can send you a floppy disk. If I'm desperate I can use the U.S. Postal Service." But even in computers, there's no free lunch. The catch is that computers have limited memories, so information has to be erased from time to time. And the Landauer principle, published in 1961, states that erasing information must use some minimum amount of energy. "Discarding information involves an unavoidable energy penality," Dr Landauer said at the Addison meeting. Landauer's principle places a theoretical limit on how energy-efficient a computer can possibly be. The principle doesn't matter much for today's real-life computers, which use much more energy than the theoretical minimum. But the principle came into play in numerous presentations at the Addison meeting, on topics ranging from black holes to the brain. For example, Christopher Fuchs of the University of North Carolina at Chapel Hill analyzed a peculiar method of erasing information with a black hole, a region of space around the remains of a collapsed star. Since the gravity of a black hole is too strong to permit anything within it to escape, a bit of information dropped into a black hole is lost forever. Does Landauer's principle apply when the information eraser is a black hole? Dr. Fuchs thinks so. A black hole swallowing a bit of information should grow in surface area, and Dr. Fuchs calculated that the amount of growth is roughly what would be expected if Landauer's principle is obeyed. David Wolpert, of the Santa Fe Institute in New Mexico, invoked Landauer's principle in explaining why the brain can only remember the past - unlike King Arthur's pal Merlin the Magician, who remembered the future. Dr. Wolpert tried to show how the psychological sense of time's one-way flow is related to the direction of time that arises from the second law of thermodynamics. Several speakers at the meeting discussed that law, which says, in essence, that disorder wins out over order. In other words, an organized system left to itself, will become disorganized. As time goes forward, eggs break but don't reassemble, building decay and machines wear out. The second law is widely regarded as among the most significant laws of physics. But it turns out that Landauer's principle is needed to save the second law from a demonic paradox. The Scottish physicist James Clerk Maxwell proposed the paradox in 1871. More than a century passed before work by Dr. Landauer and IBM colleague Charles Bennet explained it. Maxwell considered the effects of the second law in systems where something hot is separated from something cold. Open a door between a warm room and a cold room, and soon the air molecules will mix and both rooms will be the same temperature. The second law requires the mixing to create maximum disorder, or entropy. But Maxwell imagined a demon guarding the door between the rooms, opening it to allow the fast, hotter molecules into one room and closing it to keep the cold, slow molecules in the other. The demon could thus keep the rooms at different temperatures, an apparent violation of the second law. Physicists long supposed that the demon needed energy to observe the molecules and record information about their speed. The demon's energy use would create more disorder than the order he created, saving the second law. But, as Drs. Bennet and Landauer showed, computing the speeds of the molecules can be accomplished without using energy. Only when the demon resets his computer to calculate the speed of the next molecule is energy consumed, producing entropy to enforce the second law. Still, some physicists continue to wonder whether a sufficiently smart demon can outwit Dr. Landauer. Computer simulations to test new demon strategies were reported at the meeting by Andrew Rex and Ross Larsen of the University of Puget Sound in Tacoma, Wash. While the second law is still presumed to be safe, there is a way around Dr. Landauer's limit in computing - if a computer saves every intermediate result of every computation. Such a computer could then reverse every step, return to its starting place after carrying out a computation with no net use of energy, as Dr. Bennet demonstrated in 1973. But any computer keeping track of everything would need a huge memory. Ultimately, the size of such a computer would be limited by the size of the universe , Dr. Landauer pointed out. "The size of the memory is probably limited in principle, and not just by the size of your NSF (National Science Foundation) budget," he said. Of course, nobody would really try to build a computer the size of the universe. On the other hand, maybe somebody has. said Edward Fredkin of Boston University. He contends that the entire universe is just one big computer simulation. Physics, Dr. Fredkin asserts, is the constant processing of information to convert a representation of the present into a representation of the future. "Life works like a computer, thinking works like a computer and maybe physics works like a computer," he said. Whoever built this computer is not of our universe, of course. We're in the position of a pilot trainee flying from New York to Houston on a computerized 747 simulator. The computer is not in New York or Houston, but somewhere else. In the same way, says Dr. Fredkin, we live in a simulated universe running on a computer that is somewhere else. Dr. Fredkin's view isn't widely accepted. "I don't buy it," said physicist William Frensley of the University of Texas at Dallas. "I don't know if it's true or not," said physicist Seth Lloyd of Los Alamos National Laboratory in New Mexico. "I certainly doubt it's true in the sense that he (Dr. Fredkin) seems to think about it." Other speakers at the conference applied lessons from physics and computation to the mysteries of biology. Some presentations dealt with the best-known information storage system in biology - DNA, the molecule that makes up genes. DNA's genetic information is copied by RNA molecules in a cell's nucelus. The genetic information encoded in RNA molecules must then be cut up and spiced together before cells can use that information to make proteins. Thomas Schneider of the National Cancer Institute's mathenatical biology laboratory in Fredrick, Md., reported studies using information theory to understand how cells know when and where to splice. Several researchers tried to relate computational insights to the working of the human brain. Some discussed how brains represent information as patterns, others suggested ways that quantum physics could be involved in consciousness. Ordinarily, quantum physics describes the subatomic world where waves can be particles and particles can be waves, depending on how an experiment is set up. But whether quantum physics really has anything to do with consciousness is still speculation. And schemes for using curious effects of quantum physics in computing won't be showing up for sale at the Incredible Universe anytime soon. Dr. Lloyd pointed out thar quantum systems could in principle perform computations, but doing so in practice would be difficult or impossible. One quantum computing scheme, discussed by Richard Jozsa of the University of Montreal, could in theory compute complicated problems twice as fast as a standard computer. But you would only have a 50-50 chance of getting the answer to the question you asked. "You can get Two computations for the price of one, but only half of the time"," Dr. Jozsa said. Several speakers described one potential practical use for quantum physics - sending secret codes. Because quantum waves are disturbed by observation, efforts by an eavesdropper to detect a secret message can easily be detected. And an elaborate setup for sending and recieving quantum information can guarantee the ability to transmit a code no eavesdropper could possibly intercept. Claude Crepeau of the Ecole Normale Superieure in Paris suggested that the quantum coding scheme could be put to use in a bank card for use by automated teller machines. The system could be set up so that the password needed to use the card would be a perfect secret - nobody at the bank would even need to know the password. ************************************************************************ Extra box with caption Landauer's Principle vs. Maxwell's Demon Scientists have gained insights into the basic physics involved in computing -- or processing information -- by studying the connections betweeen computing and the second law of thermodynamics. That law syas that a natural system always tends to become more disordered as time passes, unless energy is imported from the system. No one has ever discovered a violation of this law. But in 1871, the scottish physicist James Clerk Maxwell proposed a paradox suggesting that a clever being, or "demon", might overcome the second law's requirements. Maxwell envisioned two chambers, one with hot air and one with cold air. According to the second law, opening the door between the chambers should cause the air molecules to mix, and bring both compartments to the same temperature. But Maxwell suggested that a demon guarding the door could open and close it selectively allowing the fast, hotter molecules to collect on one side and keeping the slow, colder molecules on the other side -- an apparent violation of the second law. Physicists long supposed that the demon needed to import energy to observe the molecules and calculate the necessary information about their speed. The use of imported energy would generate diorder outside the system to compensate for the order created inside, maintaining the requirements of the second law. But in recent years, studies by Rolf Landauer and Charles Bennet of IBM have shown that copmuting the speeds fo the molecules can be accomplished without using energy. But when the demon resets his computer to calculate the speed of the next molecule, energy is consumed, saving the second law. The notion that discarding information requires the use of energy, known as Landauer's principle, is a central consideration in a wide range of current investigations about the physics of computing. ------------------------------ End of Neuron Digest [Volume 10 Issue 19] *****************************************