neuron-request@HPLABS.HPL.HP.COM ("Neuron-Digest Moderator Peter Marvit") (06/07/90)
Neuron Digest Wednesday, 6 Jun 1990 Volume 6 : Issue 39 Today's Topics: Re: Neural Nets and forecasting Introduction to some work at NASA Symbol Train Processing submission to net: Time-Frequency Distributions & Neural Nets Networks for stereopsis Recent trends of applying NNs in digital signal processing Implementations of ART2 wanted. ART2 Source Code Final call HICSS UCLA-SFINX NN Simulator Send submissions, questions, address maintenance and requests for old issues to "neuron-request@hplabs.hp.com" or "{any backbone,uunet}!hplabs!neuron-request" Use "ftp" to get old issues from hplpm.hpl.hp.com (15.255.176.205). ------------------------------------------------------------ Subject: Re: Neural Nets and forecasting From: zt@beach.cis.ufl.edu (tang) Organization: UF CIS Department Date: 30 May 90 22:12:56 +0000 In article <3820@discg1.UUCP> ilo0005@discg1.UUCP (cherie homaee) writes: > >Has anyone used neural nets for forecasting? If so have you used any >other neural paradigm other than back-propagation? We have done some experiments with time series forecasting using back- propagation. Our results show that neural nets perform well compared with traditional methods, especially for long term forecasting. Our initial report will appear in the proceedings of the "First Workshop on Neural Networks, Auburn, 1990". See also "Neural Networks as Forecasting Experts: An Empirical Test, Proceedings of the IJCNN Meeting, Washington, 1990", by Sharda and Patil. ------------------------------ Subject: Introduction to some work at NASA From: "Eric Bobinsky" <cabobin@earth.lerc.nasa.gov> Date: 01 Jun 90 11:38:00 -0400 Salutations! I am a recent subscriber to the Digest, and I though it might be appropriate to briefly describe our neural network research program to elicit a response from other readers with similar interests and inclinations. I am with the NASA Lewis Research Center in Cleveland (yes, Cleveland!), which is NASA's lead center for satellite communications research. In the past two years we began a program with the goal of trying to apply neural network technology to the problems of enhancing the operational capabilities and service lifetimes of advanced satellite communication systems. To date, we have been working-- either directly or through university grants-- in the areas of applying neural nets to advanced satellite switching controllers, video image data compression, signal processing (particularly high-speed demodulation and decoding), and autonomous communication network control. In addition, our neural net program is tied into a much broader program in the development of advanced digital technology for high-rate modulation and coding. I'd be pleased to hear from anyone out there working in these or similar areas with whom we haven't already made acquaintance! My physical (as opposed to logical) address is: Eric Bobinsky MS 5408 (sorry, that's 54-8) Space Communications Division NASA Lewis Research Center Cleveland, Ohio 44135 Tel: 216-433-3497 FAX: 216-433-6371 ------------------------------ Subject: Symbol Train Processing From: coopere@rocky2.rockefeller.edu (Ellis D. Cooper) Organization: The Rockefeller University, NY, NY 10021 Date: 04 Jun 90 17:49:33 +0000 Symbol Train Processing Ellis D. Cooper June 1, 1990 The goal of neuroscience is to understand the fundamental principles of the brain. Communication, analysis and simulation of mental models of the underlying molecular, neuronal and network mechanisms could benefit from a standardized graphical programming language. Symbol train processing (STP) is a vivid modeling language with the added advantage of not presuming that neurons and other brain structures communicate with numbers, e.g., the activation levels of connectionism. Instead, symbol train processing assumes that brain structures communicate at all levels by emitting and absorbing sequences of symbols, only some of which might be numbers. A computer program, ChemiKine, for simulating a wide range of chemical kinetic systems using symbol train processing is available. For general symbol train processing, however, the Mathematica STP Notebook provides an object-oriented interpreter. Most neuroscientists believe that understanding the principles of the brain depends on developing theories of biological phenomena occurring on spatial scales from 0.1 meter to 1.0 thousandth of a meter in networks of neurons connected across electro-chemical synapses. Physically, a biological neural network is a dynamical system whose state space has an extremely large number of dimensions, not just because a biological neural network has a large number of synapses and neurons, but also because each synapse and neuron has many characteristic electrical potential and chemical concentration variables. Intractably complex phenomena inevitably generate diverse inquiries based on simplifying assumptions. Each inquiry hopes to provide new scientific illumination or technological applications. One technologically fruitful model of biological neural networks has been the connectionist network model. Its adequacy for neuroscientific understanding is more controversial. I am particularly interested in assumptions relating to the character and significance of spike trains. In connectionist networks the spike train is reduced to a single continuous state variable, the activation level of an abstract neurons output. A large corpus of neuroscience research is based on essentially the same abstraction of a spike train. There is also a large corpus of research in which this assumption is rejected. In fact, complex temporal patterns of action potentials are taken by many researchers to define the information produced by biological neurons. In connectionist networks the abstract neurons are passive, non-linear integrators of their inputs, whose properties are determined by coefficients in linear expressions - the weights. By contrast, biological neurons are active units with variable operating modes, including oscillator and resonator behavior. It is also implicit in the connectionist model that the individual spikes occurring in a spike train must all be identical. Biological neurons actually produce spikes of different shapes. The choice of simplifying assumptions to model a real system must be governed by criteria of verisimilitude, mathematical tractability, and computability. It can happen that it is expedient to give the latter two criteria greater emphasis at the expense of the first. This leads to arguments against using such abstract models in biology, but the successful use of ideal models in physics cannot be ignored. My purpose is to advance a new system of simplifying assumptions for model building in neuroscience which attempts to provide a superior balance between the three aforementioned criteria. STP units for building models communicate by emitting and absorbing formal symbols which can stand for spikes of different shapes, or for changes in levels of hormones, or for changes in other biologically meaningful state variables such as voltage across a membrane or current through a channel. STP units sum their simultaneous input signals and attempt to match the instantaneous sum against built-in state transition trigger symbols. STP units have intrinsic timing properties which endow them with oscillatory and resonance properties. STP units undergo both automatic and triggered transitions of state which may radically alter their signal processing properties. STP concepts were chosen specifically to apply not just at the neural network level, but also at the higher speed, smaller space scale ion channel, molecular biochemistry level, and at the lower speed, larger space scale of neuronal groups and clusters of groups, etc. The timers of STP units are easily set to random timeouts, thereby with one mechanism to model temperature at the chemical kinetics level, or the stochastic firing rates at the neuronal level. Computational neuroscience assumes that biological neural networks implement algorithms for processing information. I believe there is a theoretical need in neuroscience for a computer tool with which to simulate the brains algorithms for symbol train processing at all time scales. ------------------------------ Subject: submission to net: Time-Frequency Distributions & Neural Nets From: Don Malkoff <dmalkoff@ANDREW.dnet.ge.com> Date: Mon, 04 Jun 90 16:18:31 -0400 I am writing a review on the use of time-frequency distributions of signals as inputs to classification algorithms. The review will appear in a book "New Methods in Time-Frequency Signal Analysis" to be published by Longman & Cheshire. I am particularly (but not solely) interested in schemes where the classification mechanism is that of a neural network. I would appreciate any inputs from the net as to appropriate references. All applications are relevant. I would like to see this review be comprehensive and adequately represent the contributions of neural nets. Please reply to "dmalkoff@atl.dnet.ge.com" ____________________________________ Donald B. Malkoff General Electric Company Advanced Technology Laboratories Moorestown Corporate Center Bldg. 145-2, Route 38 Moorestown, N.J. 08057 (609) 866-6516 ------------------------------ Subject: Networks for stereopsis From: WOLPERT@VAX.OXFORD.AC.UK Organization: Physiology Department, Oxford University, UK Date: Tue, 05 Jun 90 17:33:49 +0000 I am interested in any pointers to current research/literature on neural networks for stereopsis. In particular any references to networks that solve random dot stereograms. Thanks in advance Piers Cornelissen. Reply to STEIN@UK.AC.OXFORD.VAX ------------------------------ Subject: Recent trends of applying NNs in digital signal processing From: Hazem.Abbas@QueensU.CA Date: Tue, 05 Jun 90 14:07:00 -0400 Is any body involved in the applications of neural networks in the area of digital signal processing (filter realization, adaptive filtering, image enhancement, restoration and compression). I would appreciate it if I can get acquainted with the relevant topics and bibliography as well. Actually I need that in the process of finding a research point for my Ph.D. ------------------------------ Subject: Implementations of ART2 wanted. From: RM5I%DLRVM.BITNET@CUNYVM.CUNY.EDU Date: Tue, 05 Jun 90 17:09:31 -0500 Hello, does someone have an implementation of ART2 written in a common language like Pascal or C. Thanks for any help finding this. Regards Roland Luettgens German Aerospace Research Establishment 8031 Wessling West Germany rm5i@dlrvm Bitnet ------------------------------ Subject: ART2 Source Code From: <GANKW%NUSDISCS.BITNET@CUNYVM.CUNY.EDU> Date: Wed, 06 Jun 90 17:53:00 -0800 I discovered recently that the Adaptive Resonance Theory (ART) proposed by Carpenter & Grossberg is similar in its operations to the traditional McQueen's Kmeans clustering method with coarsening and refining parameters (see ref 1). I intend to make a comparative study of these 2 methods. Is there anybody who can share with me his/her ART2 source code; or inform me how to get a copy of it? (ART1 is not suitable because my test data are\ real number vectors). I am most willing to release my findings to the network once I get the results. Thanks in advance. Reference 1. Anderberg, Cluster Analysis for Applications, Academic Press 1973. My bitnet address is : gankw@nusdiscs.bitnet Kok Wee Gan ------------------------------ Subject: Final call HICSS From: Okan K Ersoy <ersoy@ee.ecn.purdue.edu> Date: Mon, 04 Jun 90 13:28:45 -0500 FINAL CALL FOR PAPERS AND REFEREES HAWAII INTERNATIONAL CONFERENCE ON SYSTEM SCIENCES - 24 NEURAL NETWORKS AND RELATED EMERGING TECHNOLOGIES HAWAII - JANUARY 9-11, 1991 The Neural Networks Track of HICSS-24 will contain a special set of papers focusing on a broad selection of topics in the area of Neural Networks and Related Emerging Technologies. The presentations will provide a forum to discuss new advances in learning theory, associative memory, self-organization, architectures, implementations and applications. Papers are invited that may be theoretical, conceptual, tutorial or descriptive in nature. Those papers selected for presentation will appear in the Conference Proceedings which is published by the Computer Society of the IEEE. HICSS-24 is sponsored by the University of Hawaii in cooperation with the ACM, the Computer Society,and the Pacific Research Institute for Information Systems and Management (PRIISM). Submissions are solicited in: Supervised and Unsupervised Learning Issues of Complexity and Scaling Associative Memory Self-Organization Architectures Optical, Electronic and Other Novel Implementations Optimization Signal/Image Processing and Understanding Novel Applications INSTRUCTIONS FOR SUBMITTING PAPERS Manuscripts should be 22-26 typewritten, double-spaced pages in length. Do not send submissions that are significantly shorter or longer than this. Papers must not have been previously presented or published, nor currently submitted for journal publication. Each manuscript will be put through a rigorous refereeing process. Manuscripts should have a title page that includes the title of the paper, full name of its author(s), affiliations(s), complete physical and electronic address(es), telephone number(s) and a 300-word abstract of the paper. DEADLINES Six copies of the manuscript are due by June 25, 1990. Notification of accepted papers by September 1, 1990. Accepted manuscripts, camera-ready, are due by October 3, 1990. SEND SUBMISSIONS AND QUESTIONS TO O. K. Ersoy Purdue University School of Electrical Engineering W. Lafayette, IN 47907 (317) 494-6162 E-Mail: ersoy@ee.ecn.purdue.edu ------------------------------ Subject: UCLA-SFINX NN Simulator From: Edmond Mesrobian <edmond@CS.UCLA.EDU> Date: Mon, 04 Jun 90 13:25:47 -0700 Recently, there was a posting concerning SFINX. The information was a bit incorrect. To obtain the simulator, one must first sign a license agreement. FTP instructions will then be sent to licensee. More information concerning the simulator is presetned below. hope this helps, Edmond Mesrobian UCLA Machine Perception Lab 3531 Boelter Hall Los Angeles, CA 90024 ============================================================================ UCLA-SFINX ( Structure and Function In Neural connec- tions) is an interactive neural network simulation environment designed to provide the investigative tools for studying the behavior of various neural structures. It was designed to easily express and simulate the highly regular patterns often found in large networks, but it is also general enough to model parallel systems of arbitrary interconnectivity. UCLA-SFINX is not based on any single neural network para- digm such as Backward Error Propagation (BEP) but rather enables users to simulate a wide variety of neural network models. UCLA- SFINX has been used to simulate neural networks for the segmenta- tion of images using textural cues, architectures for color and lightness constancy, script character recognition using BEP and others. It is all written in C, includes an X11 interface for visual- izing simulation results (8 bit displays), and it has been ported to HP 9000 320/350 workstations running HP-UX, Sun workstations running SUNOS 3.5, IBM RT workstations running BSD 4.3, Ardent Titan workstations running Ardent UNIX Release 2.0, and VAX 8200's running Ultrix 2.2-1. To get UCLA-SFINX source code and document- ation (in LaTeX format) follow the instructions below: 1. To obtain UCLA-SFINX via the Internet: Sign and return the enclosed UCLA-SFINX License Agreement to the address below. We will send you a copy of the signed license agreement along with instructions on how to FTP a copy of UCLA-SFINX. If you have a PostScript printer, you should be able to produce your own copy of the manual. If you wish to obtain a hardcopy of the manual, return a check for $30 along with the license. 2. To obtain UCLA-SFINX on tape: Sign and return the enclosed UCLA-SFINX License Agreement to the address below. Return a check for $100 dollars along with the license, for a hardcopy of the manual and a copy of UCLA-SFINX on 1/4 inch cartridge tape (in tar format) read- able by a Sun 3 workstation. We will also send you a copy of the signed license agreement. Checks should be made payable to the Regents of the Univer- sity of California. If you have questions regarding any of the information discussed above send electronic mail to sfinx@retina.cs.ucla.edu or US mail to: UCLA Machine Perception Laboratory, Computer Science Department, 3532 Boelter Hall, Los Angeles, CA. 90024, USA. >>>>>>>>>>>>>>>>>>>>>>>>>>>>> cut here for license <<<<<<<<<<<<<<<<<<<<<<<<< THE REGENTS OF THE UNIVERSITY OF CALIFORNIA LOS ANGELES CAMPUS UCLA-SFINX LICENSE AGREEMENT This Agreement is entered into this___________of ____________________, 199__, by and between THE REGENTS OF THE UNIVERSITY OF CALIFORNIA, a California corporation, hereinafter called "University", and ________________________________________ _____________________________________, hereinafter called "Licensee." This Agreement is made with reference to the follow- ing: 1. DEFINITION "UCLA-SFINX" is a set of copyrighted, source code computer programs and any future modifications thereof delivered by University to Licensee, and any accompanying documentation provided by University. UCLA-SFINX is a general purpose software system for the development and evaluation of con- nectionist models. UCLA-SFINX is written for and operates on UNIX systems. 2. GRANT OF RIGHTS A. University grants to Licensee and Licensee accepts a non-exclusive, non-transferable license to use UCLA- SFINX solely for Licensee's non-commercial purposes. B. Such use may include the making of sufficient copies of UCLA-SFINX for the reasonable purposes of Licensee hereunder. All copies of UCLA-SFINX made by Licensee, in whole or in part, regardless of the form in which the Licensee may subsequently use it, and regardless of any modification which the Licensee may subsequently make to it are the property of University and no title to or ownership of such materials are transferred to Licensee hereunder. Licensee shall include on any such copies labels containing the name UCLA-SFINX, the University's copyright notice, and any other proprietary or restrictive notices appearing on the la- bel of the copy of UCLA-SFINX furnished to Licensee by University. 1 C. Such use may include the modification of UCLA-SFINX by Licensee. Such modified versions of UCLA-SFINX shall remain the property of University. D. Such use shall not include further distribution, or any action which may be construed as selling or licensing UCLA-SFINX to any person or entity. 3. ACKNOWLEDGMENT A. Licensee acknowledges that UCLA-SFINX has been developed for research purposes only. B. Licensee shall require its employees and students to acknowledge in writing their use of UCLA-SFINX when re- porting any research resulting from such use. The fol- lowing notice should be used: "UCLA-SFINX from UCLA MACHINE PERCEPTION LABORATORY." 4. WARRANTIES AND INDEMNIFICATION A. University warrants that is is the owner of all right, title, and interest in and to UCLA-SFINX, including all copyright pertaining thereto and subsisting therein. B. UCLA-SFINX is licensed "AS IS," and University dis- claims all warranties, express and implied, including but not limited to, the implied warranties of merchan- tability and fitness for a particular purpose. In no event will University be liable for any business ex- pense, machine down time, loss of profits, any inciden- tal, special, exemplary or consequential damages, or any claims or demands brought against Licensee. The entire risk as to the quality and performance of UCLA- SFINX is with Licensee. C. Licensee agrees to indemnify, defend, hold harmless, and defend University, its officers, employees and agents, against any and all claims, suits, losses, dam- ages, costs, fees, and expenses resulting from or aris- ing out of any use of UCLA-SFINX by Licensee. 5. TECHNICAL SUPPORT AND FEEDBACK A. University shall have no obligation to install, sup- port, maintain, or correct any defects in UCLA-SFINX. 2 B. Licensee agrees to notify University of any errors, functional problems, and any defects in performance discovered in UCLA-SFINX and of any fixes made by Licensee. Such notice will contain a full description of the problem, indicating in what circumstances it originated, and how it manifested itself. Technical matters and errors discovered in UCLA-SFINX may be com- municated as provided in Article 9 below or via elec- tronic mail to: sfinx@retina.cs.ucla.edu. 6. TERM AND TERMINATION A. The term of this Agreement is perpetual and shall be effective from the date of its signing by duly author- ized official of Licensee. B. Any failure of Licensee to comply with all terms and conditions of this Agreement shall result in its im- mediate termination. 7. SEVERABILITY If any of the provisions or portions of this Agreement are invalid under any applicable statute or rule of law, they are to the extent of such invalidity severable and shall not affect any other provision of this Agreement. 8. APPLICABLE LAW This Agreement shall be governed by the laws of the State of California. 9. NOTICE A. Any notice under this Agreement shall be in writing and mailed to the appropriate address given below: To University regarding this Agreement: The Regents of the University of California Office of Contract and Grant Administration University of California, Los Angeles 405 Hilgard Avenue Los Angeles, California 90024-1406 Attention: Dr. Enrique Riveros-Schafer 3 B. To University regarding technical matters: UCLA Machine Perception Laboratory 3532 Boelter Hall Computer Science Department Los Angeles, California 90024 Attention: Prof. Josef Skrzypek C. To Licensee: ____________________________________________________ ____________________________________________________ ____________________________________________________ ____________________________________________________ ____________________________________________________ 10. ENTIRETY This Agreement supersedes any previous communication and, when signed by both parties, constitutes the complete under- standing of the parties. No modification or waiver of any provisions hereof shall be valid unless in writing and signed by both parties. IN WITNESS THEREOF, the parties here to have caused this Agree- ment to be executed. LICENSEE THE REGENTS OF THE UNIVERSITY OF CALIFORNIA By: _______________________________ By: ______________________________ NAME: _______________________________ Wade A. Bunting, Ph.D. Title: _______________________________ Intellectual Property Officer Date: _______________________________ Date: ____________________________ 4 >>>>>>>>>>>>>>>>>>>>>>>>>>>>> cut here for license <<<<<<<<<<<<<<<<<<<<<<<<< ------------------------------ End of Neuron Digest [Volume 6 Issue 39] ****************************************