From sml@esesparc2.essex.ac.uk Sun Sep 1 13:41:52 1996 Received: from lucy.cs.wisc.edu (lucy.cs.wisc.edu [128.105.2.11]) by sea.cs.wisc.edu (8.6.12/8.6.12) with ESMTP id NAA01112 for ; Sun, 1 Sep 1996 13:41:33 -0500 Received: from TELNET-1.SRV.CS.CMU.EDU (TELNET-1.SRV.CS.CMU.EDU [128.2.254.108]) by lucy.cs.wisc.edu (8.6.12/8.6.12) with SMTP id NAA18745 for ; Sun, 1 Sep 1996 13:41:31 -0500 Received: from TELNET-1.SRV.CS.CMU.EDU by telnet-1.srv.cs.CMU.EDU id ab27162; 1 Sep 96 12:34:20 EDT Received: from DST.BOLTZ.CS.CMU.EDU by TELNET-1.SRV.CS.CMU.EDU id ab27151; 1 Sep 96 12:07:25 EDT Received: from DST.BOLTZ.CS.CMU.EDU by DST.BOLTZ.CS.CMU.EDU id aa00903; 1 Sep 96 12:06:53 EDT Received: from CS.CMU.EDU by B.GP.CS.CMU.EDU id aa14581; 1 Sep 96 7:40:54 EDT Received: from [155.245.10.186] by CS.CMU.EDU id aa29518; 1 Sep 96 7:40:33 EDT Received: from esesparc2.essex.ac.uk by seralph6.essex.ac.uk; (5.65v3.2/1.1.8.2/14Feb95-1154AM) id AA24637; Tue, 27 Aug 1996 18:29:23 +0100 Date: Tue, 27 Aug 1996 18:29:24 +0100 (BST) From: Lucas S M To: Connectionists%cs.cmu.edu@seralph6.essex.ac.uk Subject: papers available on high performance OCR Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Summary: The following two papers discuss recent work on applying scanning n-tuple classifiers to handwritten OCR. The first is a journal paper which gives some background and all the technical details. The second is a paper for a forthcoming conference which includes more up-to-date results and more detailed timing analysis. The main feature of the method is the incredible speed. If we ignore the pre-processing time, we can train the system at a rate of over 20,000 character images per second, and recognise about 1,200 characters per second, on a humble 66mhz Pentium PC. If we include pre-processing time, then we can still train (recognise) 500 (200) chars per second. The fast training and recognition speeds allow the system parameters to be optimised very quickly. Best accuracy reported is 98.3% on the CEDAR hand-written digit test set. This is not quite at good as the best reported in the literature for this data (98.9%, to the best of our knowledge), but offers a significant speed advantage. The following two papers discuss recent work on a simple unified approach to evolving ALL aspects of a neural network, including its learning algorithm (if any). The first uses a grammar based chromosome, the second uses a set-based chromosome. The latter approach appears particularly promising as a method of part designing/ part evolving neural networks. ----------------------------------------------------------------------- Title: Statistical syntactic Methods for high performance OCR Author: S.M. Lucas and A. Amiri Date: Feb 1996 In: IEE Proceedings on Vision, Image and Signal Processing, vol 143, pp 23 -- 30. This paper is available through my web page: http://esewww.essex.ac.uk/~sml or via anonymous ftp: ftp tarifa.essex.ac.uk cd /images/sml/reports get ieevisp.ps ----------------------------------------------------------------------- Title: Improving scanning n-tuple classifiers by pre-transforming training data Author: S.M. Lucas Date: Sep 1996 In: (to appear) Proceedings International Workshop on Frontiers in Handwriting Recognition, University of Essex 1996 pp 143 -- 146 This paper is available through my web page: http://esewww.essex.ac.uk/~sml or via anonymous ftp: ftp tarifa.essex.ac.uk cd /images/sml/reports get iwfhr96.ps ----------------------------------------------------------------------- Comments and criticisms welcome. Simon Lucas ------------------------------------------------ Dr. Simon Lucas Department of Electronic Systems Engineering University of Essex Colchester CO4 3SQ United Kingdom http://esewww.essex.ac.uk/~sml Tel: (+44) 1206 872935 Fax: (+44) 1206 872900 Email: sml@essex.ac.uk secretary: Mrs Wendy Ryder (+44) 1206 872437 ------------------------------------------------- From denni@bordeaux.cse.ogi.edu Sun Sep 1 13:41:53 1996 Received: from lucy.cs.wisc.edu (lucy.cs.wisc.edu [128.105.2.11]) by sea.cs.wisc.edu (8.6.12/8.6.12) with ESMTP id NAA01110 for ; Sun, 1 Sep 1996 13:41:31 -0500 Received: from TELNET-1.SRV.CS.CMU.EDU (TELNET-1.SRV.CS.CMU.EDU [128.2.254.108]) by lucy.cs.wisc.edu (8.6.12/8.6.12) with SMTP id NAA18743 for ; Sun, 1 Sep 1996 13:41:29 -0500 Received: from TELNET-1.SRV.CS.CMU.EDU by telnet-1.srv.cs.CMU.EDU id aa27162; 1 Sep 96 12:33:11 EDT Received: from DST.BOLTZ.CS.CMU.EDU by TELNET-1.SRV.CS.CMU.EDU id aa27151; 1 Sep 96 12:07:19 EDT Received: from DST.BOLTZ.CS.CMU.EDU by DST.BOLTZ.CS.CMU.EDU id aa00898; 1 Sep 96 12:06:44 EDT Received: from RI.CMU.EDU by B.GP.CS.CMU.EDU id aa10916; 1 Sep 96 1:09:19 EDT Received: from cse.ogi.edu by RI.CMU.EDU id aa05083; 1 Sep 96 1:08:39 EDT Received: from bordeaux.cse.ogi.edu by church.cse.ogi.edu with smtp (Smail3.1.29.1 #4) id m0ux4lw-000KrlC; Sat, 31 Aug 96 22:08 PDT Received: by bordeaux.cse.ogi.edu (NX5.67e/NX3.0S) id AA20633; Sat, 31 Aug 96 22:08:36 -0700 Message-Id: <9609010508.AA20633@bordeaux.cse.ogi.edu> Content-Type: text/plain Mime-Version: 1.0 (NeXT Mail 3.3 v118.2) Received: by NeXT.Mailer (1.118.2) From: Thorsteinn Rognvaldsson Date: Sat, 31 Aug 96 22:08:34 -0700 To: Connectionists@cs.cmu.edu Subject: Smoothing Regularizers for PBF NN New tech report available: SMOOTHING REGULARIZERS FOR PROJECTIVE BASIS FUNCTION NETWORKS By: JOHN E. MOODY & THORSTEINN S. ROGNVALDSSON Dept. of Computer Science and Engineering Oregon Graduate Institute of Science and Technology P.O. Box 91000 Portland, Oregon 97291-1000, U.S.A. Emails: moody@cse.ogi.edu denni@cse.ogi.edu (Direct correspondence to Prof. Moody) --------- Abstract: Smoothing regularizers for radial basis functions have been studied extensively, but no general smoothing regularizers for PROJECTIVE BASIS FUNCTIONS (PBFs), such as the widely-used sigmoidal PBFs, have heretofore been proposed. We derive new classes of algebraically-simple m:th-order smoothing regularizers for networks of projective basis functions. Our simple algebraic forms enable the direct enforcement of smoothness without the need for e.g. costly Monte Carlo integrations of the smoothness functional. We show that our regularizers are highly correlated with the values of standard smoothness functionals, and thus suitable for enforcing smoothness constraints onto PBF networks. The regularizers are tested on illustrative sample problems and compared to quadratic weight decay. The new regularizers are shown to yield better generalization errors than weight decay when the implicit assumptions in the latter are wrong. Unlike weight decay, the new regularizers distinguish between the roles of the input and output weights and capture the interactions between them. -------------------------------------------------- Instructions for retrieving your own personal copy: WWW: http://www.cse.ogi.edu/~denni/publications.html FTP: % ftp neural.cse.ogi.edu (username=anonymous, password=your email) > cd pub/neural/papers/ > get moodyRogn96.smooth_long.ps.Z > quit % uncompress moodyRogn96.smooth_long.ps.Z % lpr moodyRogn96.smooth_long.ps (assumes you have a UNIX system) From maggini@sun1.ing.unisi.it Mon Sep 2 12:32:25 1996 Received: from lucy.cs.wisc.edu (lucy.cs.wisc.edu [128.105.2.11]) by sea.cs.wisc.edu (8.6.12/8.6.12) with ESMTP id MAA09484 for ; Mon, 2 Sep 1996 12:32:17 -0500 Received: from TELNET-1.SRV.CS.CMU.EDU (TELNET-1.SRV.CS.CMU.EDU [128.2.254.108]) by lucy.cs.wisc.edu (8.6.12/8.6.12) with SMTP id MAA25575 for ; Mon, 2 Sep 1996 12:32:14 -0500 Received: from TELNET-1.SRV.CS.CMU.EDU by telnet-1.srv.cs.CMU.EDU id aa29421; 2 Sep 96 8:58:23 EDT Received: from DST.BOLTZ.CS.CMU.EDU by TELNET-1.SRV.CS.CMU.EDU id aa29418; 2 Sep 96 8:47:13 EDT Received: from DST.BOLTZ.CS.CMU.EDU by DST.BOLTZ.CS.CMU.EDU id aa01682; 2 Sep 96 8:46:53 EDT Received: from RI.CMU.EDU by B.GP.CS.CMU.EDU id aa29249; 2 Sep 96 3:26:57 EDT Received: from alpha1.ing.unisi.it by RI.CMU.EDU id aa08127; 2 Sep 96 3:26:27 EDT Received: from sun1.ing.unisi.it by alpha1.ing.unisi.it; (5.65v3.2/1.1.8.2/27Aug96-1049AM) id AA11372; Mon, 2 Sep 1996 09:26:54 +0100 Received: from ultra1 by sun1 (SMI-8.6/SMI-SVR4) id JAA00740; Mon, 2 Sep 1996 09:24:19 +0200 Received: by ultra1 (SMI-8.6/SMI-SVR4) id JAA08804; Mon, 2 Sep 1996 09:23:09 +0200 Date: Mon, 2 Sep 1996 09:23:09 +0200 From: Marco Maggini Message-Id: <199609020723.JAA08804@ultra1> To: Connectionists@cs.cmu.edu Subject: NIPS'96 Postconference Workshop Reply-To: marco@mcculloch.ing.unifi.it Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Content-Md5: x+uYPkPbUy339M+ONMdTtA== ================================================================= CALL FOR PAPERS NIPS'96 Postconference Workshop ----------------------------------------------------------------- ANNs and Continuous Optimization: Local Minima, Sub-optimal Solutions, and Computational Complexity ----------------------------------------------------------------- Snowmass (Aspen), Colorado USA Fri Dec 6th, 1996 M. Gori M. Protasi Universita' di Siena Universita' Tor Vergata (Roma) Universita' di Firenze protasi@utovrm.it marco@mcculloch.ing.unifi.it http://www-dsi.ing.unifi.it/neural Most ANNs used for either learning or problem solving (e.g. Backprop nets and analog Hopfield nets) rely on continuous optimization. The elegance and generality of this approach, however, seems also to represent the main source of troubles that typically arise when approaching complex problems. Most of the times this gives rise to a sort of suspiciousness concerning the actual chance to discover an optimal solution under reasonable computational constraints. The computational complexity of the problem at hand seems to appear in terms of local minima and numerical problems of the chosen optimization algorithm. While most practitioners use to accept without reluctance the flavor of suspiciousness and use to be proud of their eventual experimental achievements, most theoreticians are instead quite skeptical. Obviously, the success of ANNs for either learning and problem solving is often related to the problem at hand and, therefore, one can expect an excellent behavior for a class of problems, while can raise serious suspects about the solution of others. To the best of our knowledge, however, so far, this intuitive idea has no satisfactory theoretical explanation. Basically, there is neither theory to support naturally the intuitive concept of suspiciousness, nor theory to relate this concept to computational complexity. The study of the complexity of algorithms has been essentially performed on discrete structures and an impressive theory has been developed in the last two decades. On the other hand optimization theory has a twofold face: discrete optimization and continuous optimization. Actually there are some important approaches of the computational complexity theory that were proposed for the continuous cases, for instance, the information based-complexity (Traub) and real Turing Machines (Blum-Shub-Smale). These approaches can be fruitfully applied to problems arising in continuous optimization but, generally speaking, the formal study of the efficiency of algorithms and problems has received much more attention in the discrete environment, where the theory can be easily used and error and precision problems are not present. =========================================== DISCUSSION POINTS FOR WORKSHOP PARTICIPANTS =========================================== Taking into account this framework, a fascinating area, that we believe deserves a careful study, concerns the relationship between the emergence of sub-optimal solutions in continuous optimization and the corresponding computational complexity of the problem at hand. More specifically,there are a number of very intriguing open questions: - Is there a relationship between the complexity of algorithms in the continuous and discrete settings for solving the same problem? - Can we deduce bounds on the complexity of discrete algorithms from the study of the properties of continuous ones and vice-versa? - Some loading problems are intuitively easily solvable, while others are considered hard. Are there links between the presence of local minima and the computational complexity of the problem at hand? - What is the impact of approximate solutions on the complexity? (e.g. learning is inherently an approximate process whose complexity is often studied in the framework of theories like PAC) ========== OBJECTIVES ========== The aim of the workshop is not to reach some definite points, but to stimulate a starting discussion, and to put on the table some of the most important themes that we hope could be extensively explored in the future. We also expect that the study of the interplay between continuous and discrete versions of a problem can be very fruitful for both the approaches. Since, until now, this interplay has been rarely explored, the workshop is likely to stimulate different point of view; people working on discrete optimization are in fact likely not to be expert on the continuous side and vice-versa. ========================================= SUBMISSION OF WORKSHOP EXTENDED ABSTRACTS ========================================= If you would like to contribute, please send an abstract or extended summary to: Marco Gori Facolta' di Ingegneria Universita' di Siena Via Roma, 56 53100 Siena (Italy) Fax: +39 577 26.36.02 Electronic submission: Send manuscripts in postscript format at marco@mcculloch.ing.unifi.it Important Dates: Submission of abstract deadline: 30 September, 1996 Notification of acceptance: 21 October, 1996 Final paper to be sent by: 4 November, 1996 For the format of papers, the usual NIPS style file should be used with up to 16 pages allowed. Workshop notes will be produced. NIPS style files are available at http://www.cs.cmu.edu/afs/cs/project/cnbc/nips/formatting/nips.sty http://www.cs.cmu.edu/afs/cs/project/cnbc/nips/formatting/nips.tex http://www.cs.cmu.edu/afs/cs/project/cnbc/nips/formatting/nips.ps Please contact the workshop organizers for further information, or consult the NIPS WWW home page: http://www.cs.cmu.edu/afs/cs.cmu.edu/Web/Groups/NIPS/ From pfbaldi@cco.caltech.edu Tue Sep 3 00:31:49 1996 Received: from lucy.cs.wisc.edu (lucy.cs.wisc.edu [128.105.2.11]) by sea.cs.wisc.edu (8.6.12/8.6.12) with ESMTP id AAA14556 for ; Tue, 3 Sep 1996 00:31:38 -0500 Received: from TELNET-1.SRV.CS.CMU.EDU (TELNET-1.SRV.CS.CMU.EDU [128.2.254.108]) by lucy.cs.wisc.edu (8.6.12/8.6.12) with SMTP id AAA00750 for ; Tue, 3 Sep 1996 00:31:36 -0500 Received: from TELNET-1.SRV.CS.CMU.EDU by telnet-1.srv.cs.CMU.EDU id aa01400; 2 Sep 96 22:56:34 EDT Received: from DST.BOLTZ.CS.CMU.EDU by TELNET-1.SRV.CS.CMU.EDU id aa01397; 2 Sep 96 22:45:00 EDT Received: from DST.BOLTZ.CS.CMU.EDU by DST.BOLTZ.CS.CMU.EDU id aa02028; 2 Sep 96 22:44:50 EDT Received: from EDRC.CMU.EDU by B.GP.CS.CMU.EDU id aa11531; 2 Sep 96 20:01:21 EDT Received: from accord.cco.caltech.edu by EDRC.CMU.EDU id aa08295; 2 Sep 96 20:00:43 EDT Received: from localhost by accord.cco.caltech.edu with SMTP (8.6.12/DEI:4.45) id QAA00180; Mon, 2 Sep 1996 16:59:00 -0700 Date: Mon, 2 Sep 1996 16:58:40 -0700 (PDT) From: Pierre Baldi X-Sender: pfbaldi@accord To: Connectionists@cs.cmu.edu cc: Pierre Baldi Subject: TR available: Bayesian Methods and Compartmental Modeling Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII FTP-host: archive.cis.ohio-state.edu FTP-filename: /pub/neuroprose/baldi.comp.tar.Z The file baldi.comp.tar.Z is now available for copying from the Neuroprose repository: ON THE USE OF BAYESIAN METHODS FOR EVALUATING COMPARTMENTAL NEURAL MODELS (40 pages = 35 pages + 5 figures) (one figure is in color but should print OK in black and white) P. Baldi, M. C. Vanier, and J. M. Bower Department of Computation and Neural Systems Caltech ABSTRACT: In this TR, we provide a tutorial on Bayesian methods for neurobiologists, as well an application of the methods to compartmental modeling. We first derive prior and likelihood functions for compartmental neural models and for spike trains. We then apply the full Bayesian inference machinery to parameter estimation, and model comparison in the case of simple classes of compartmental models, with three and four conductances. We also perform class comparison by approximating integrals over the entire parameter space. Advantages and drawbacks are discussed. Sorry-no hard copies available. From surmeli@pmt11.et.Uni-Magdeburg.DE Tue Sep 3 12:41:46 1996 Received: from lucy.cs.wisc.edu (lucy.cs.wisc.edu [128.105.2.11]) by sea.cs.wisc.edu (8.6.12/8.6.12) with ESMTP id MAA22282 for ; Tue, 3 Sep 1996 12:41:35 -0500 Received: from TELNET-1.SRV.CS.CMU.EDU (TELNET-1.SRV.CS.CMU.EDU [128.2.254.108]) by lucy.cs.wisc.edu (8.6.12/8.6.12) with SMTP id MAA07197 for ; Tue, 3 Sep 1996 12:41:29 -0500 Received: from TELNET-1.SRV.CS.CMU.EDU by telnet-1.srv.cs.CMU.EDU id aa02135; 3 Sep 96 4:14:41 EDT Received: from DST.BOLTZ.CS.CMU.EDU by TELNET-1.SRV.CS.CMU.EDU id aa02133; 3 Sep 96 4:05:47 EDT Received: from DST.BOLTZ.CS.CMU.EDU by DST.BOLTZ.CS.CMU.EDU id aa02487; 3 Sep 96 4:04:55 EDT Received: from CS.CMU.EDU by B.GP.CS.CMU.EDU id aa16631; 3 Sep 96 3:03:47 EDT Received: from pmt11.et.Uni-Magdeburg.DE by CS.CMU.EDU id aa09573; 3 Sep 96 3:03:19 EDT Received: from pmt22 by pmt11.et.uni-magdeburg.de with smtp (Smail3.1.28.1 #3) id m0uxpNk-0009qlC; Tue, 3 Sep 96 08:54 MET DST Sender: surmeli@DST.BOLTZ.CS.CMU.EDU Message-ID: <322BD693.3D70@ipe.et.uni-magdeburg.de> Date: Tue, 03 Sep 1996 08:56:19 +0200 From: Dimitrij Surmeli X-Mailer: Mozilla 3.0 (X11; I; SunOS 5.4 sun4m) MIME-Version: 1.0 Newsgroups: comp.ai.neural-nets,de.markt.arbeit.angebote,sci.research.postdoc,bln.sci.misc,bln.jobs CC: connectionists@cs.cmu.edu, neuron@CATTELL.PSYCH.UPENN.EDU, cogpsy@cogsci.soton.ac.uk, laws@computists.com, surmeli@pmt11.et.Uni-Magdeburg.DE, michaelis@pmt11.et.Uni-Magdeburg.DE Subject: Job: GRA in neural nets for control; Magdeburg, Germany Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Job announcement: The Institute of Measurement and Electronics of the Otto-von-Guericke-University, Magdeburg, Germany has an opening for a Research Assistant in the 'Innovationskolleg ADAMES' as of 1 October 1996. The project is investigating distributed neural network applications for ADAptive MEchanical Systems (ADAMES), ie signal identification, data compression and control systems. This is one area in a multi-disciplinary project involving actively deformable mechanical systems. Desired qualifications include a finished BSc/MSc, experience in neural network applications, control theory, image processing, programming (all in Unix and Win). Helpful: Matlab experience, hardware design, CNAPS neurocomputer experience, analog-digital-analog conversion Working language German. Compensation depending on qualification on the BAT II-O scale. Position suitable to engage in research leading to PhD. Informal inquiries re: ADAMES to surmeli@ipe.et.uni-magdeburg.de Formal application including CV, cover letter, etc to: Prof. B. Michaelis otto-von-Guericke Universitaet Madgeburg Fakultaet fuer Elektrotechnik Institut fuer Prozessmesstechnik und Elektronik Am Kroekentor 2 39106 Magdeburg Germany tel. +49 391 671 4645 fax +49 391 561 6368 email michaelis@ipe.et.uni-magdeburg.de http://pmt05.et.uni-magdeburg.de/TI/TI.html -- Dimitrij Surmeli surmeli@ipe.et.uni-magdeburg.de Anybody got a good name for a neurocomputer CNAPS 512? From payman@u.washington.edu Wed Sep 4 03:04:06 1996 Received: from lucy.cs.wisc.edu (lucy.cs.wisc.edu [128.105.2.11]) by sea.cs.wisc.edu (8.6.12/8.6.12) with ESMTP id DAA08439 for ; Wed, 4 Sep 1996 03:03:53 -0500 Received: from TELNET-1.SRV.CS.CMU.EDU (TELNET-1.SRV.CS.CMU.EDU [128.2.254.108]) by lucy.cs.wisc.edu (8.6.12/8.6.12) with SMTP id DAA18352 for ; Wed, 4 Sep 1996 03:03:51 -0500 Received: from TELNET-1.SRV.CS.CMU.EDU by telnet-1.srv.cs.CMU.EDU id aa03740; 4 Sep 96 0:05:21 EDT Received: from DST.BOLTZ.CS.CMU.EDU by TELNET-1.SRV.CS.CMU.EDU id aa03738; 3 Sep 96 23:53:15 EDT Received: from DST.BOLTZ.CS.CMU.EDU by DST.BOLTZ.CS.CMU.EDU id aa03247; 3 Sep 96 23:52:39 EDT Received: from RI.CMU.EDU by B.GP.CS.CMU.EDU id aa07833; 3 Sep 96 21:43:54 EDT Received: from saul3.u.washington.edu by RI.CMU.EDU id aa17293; 3 Sep 96 21:43:21 EDT Received: by saul3.u.washington.edu (5.65+UW96.08/UW-NDC Revision: 2.33 ) id AA28874; Tue, 3 Sep 96 18:43:19 -0700 Date: Tue, 3 Sep 96 18:43:19 -0700 From: Payman Arabshahi Message-Id: <9609040143.AA28874@saul3.u.washington.edu> X-Sender: payman@saul3.u.washington.edu To: connectionists@cs.cmu.edu Subject: CFP: 1997 Computational Intelligence in Financial Eng. CIFEr IEEE/IAFE 1997 $$$$$$$$$$$ $$$$$$ $$$$$$$$$$$ $$$$$$$$$$ $$$$$$$$$$$ $$$$$$ $$$$$$$$$$$ $$$$$$$$$$ $$$$ $$ $$$$ $$$$ $$$ $$$ $$$$ $$$$ $$$$$$$ $$$$$$ $$$$$$$$$$ $$$$ $$$$ $$$$$$$ $$$$$$ $$$$$$$$$$ $$$$ $$ $$$$ $$$$ $$$ $$$ $$$ $$$$$$$$$$$ $$$$$$ $$$$ $$$$$$$$$$ $$$ $$$$$$$$$$$ $$$$$$ $$$$ $$$$$$$$$$ $$$ Visit us on the web at http://www.ieee.org/nnc/cifer97 ------------------------------------ ------------------------------------ Call for Papers Conference Topics Conference on Computational ------------------------------------ Intelligence for Financial Engineering Topics in which papers, panel sessions, and tutorial proposals are (CIFEr) invited include, but are not limited to, the following: Crowne Plaza Manhattan, New York City Financial Engineering Applications: March 23-25, 1997 * Risk Management * Pricing of Structured Sponsors: Securities The IEEE Neural Networks Council, * Asset Allocation The International Association of * Trading Systems Financial Engineers * Forecasting * Hedging Strategies The IEEE/IAFE CIFEr Conference is * Risk Arbitrage the third annual collaboration * Exotic Options between the professional engineering and financial communities, and is Computer & Engineering Applications one of the leading forums for new & Models: technologies and applications in the intersection of computational * Neural Networks intelligence and financial * Probabilistic Modeling/Inference engineering. Intelligent * Fuzzy Systems and Rough Sets computational systems have become * Genetic and Dynamic Optimization indispensable in virtually all * Intelligent Trading Agents financial applications, from * Trading Room Simulation portfolio selection to proprietary * Time Series Analysis trading to risk management. * Non-linear Dynamics ------------------------------------------------------------------------------ Instructions for Authors, Special Sessions, Tutorials, & Exhibits ------------------------------------------------------------------------------ All summaries and proposals for tutorials, panels and special sessions must be received by the conference Secretariat at Meeting Management by November 15, 1996. Our intentions are to publish a book with the best selection of papers accepted. Authors (For Conference Oral Sessions) One copy of the Extended Summary (not exceeding four pages of 8.5 inch by 11 inch size) must be received by Meeting Management by November 15, 1996. Centered at the top of the first page should be the paper's complete title, author name(s), affiliation(s), and mailing addresses(es). Fonts no smaller than 10 pt should be used. Papers must report original work that has not been published previously, and is not under consideration for publication elsewhere. In the letter accompanying the submission, the following information should be included: * Topic(s) * Full title of paper * Corresponding Author's name * Mailing address * Telephone and fax * E-mail (if available) * Presenter (If different from corresponding author, please provide name, mailing address, etc.) Authors will be notified of acceptance of the Extended Summary by January 10, 1997. Complete papers (not exceeding seven pages of 8.5 inch by 11 inch size) will be due by February 14, 1997, and will be published in the conference proceedings. ---------------------------------------------------------------------------- Special Sessions A limited number of special sessions will address subjects within the topical scope of the conference. Each special session will consist of from four to six papers on a specific topic. Proposals for special sessions will be submitted by the session organizer and should include: * Topic(s) * Title of Special Session * Name, address, phone, fax, and email of the Session Organizer * List of paper titles with authors' names and addresses * One page of summaries of all papers Notification of acceptance of special session proposals will be on January 10, 1997. If a proposal for a special session is accepted, the authors will be required to submit a camera ready copy of their paper for the conference proceedings by February 14, 1997. ---------------------------------------------------------------------------- Panel Proposals Proposals for panels addressing topics within the technical scope of the conference will be considered. Panel organizers should describe, in two pages or less, the objective of the panel and the topic(s) to be addressed. Panel sessions should be interactive with panel members and the audience and should not be a sequence of paper presentations by the panel members. The participants in the panel should be identified. No papers will be published from panel activities. Notification of acceptance of panel session proposals will be on January 10, 1997. ---------------------------------------------------------------------------- Tutorial Proposals Proposals for tutorials addressing subjects within the topical scope of the conference will be considered. Proposals for tutorials should describe, in two pages or less, the objective of the tutorial and the topic(s) to be addressed. A detailed syllabus of the course contents should also be included. Most tutorials will be four hours, although proposals for longer tutorials will also be considered. Notification of acceptance of tutorial proposals will be on January 10, 1997. ---------------------------------------------------------------------------- Exhibit Information Businesses with activities related to financial engineering, including software & hardware vendors, publishers and academic institutions, are invited to participate in CIFEr's exhibits. Further information about the exhibits can be obtained from the CIFEr-secretariat, Barbara Klemm. ---------------------------------------------------------------------------- Contact Information Sponsors More information on registration and Sponsorship for CIFEr'97 the program will be provided as soon is being provided by the IAFE as it becomes available. For further (International Association of details, please contact Financial Engineers) and the IEEE Neural Networks Council. The IEEE Barbara Klemm (Institute of Electrical and CIFEr'97 Secretariat Electronics Engineers) is the Meeting Management world's largest engineering and IEEE/IAFE Computational Intelligence computer science professional for Financial Engineering non-profit association and sponsors 2603 Main Street, Suite # 690 hundreds of technical conferences Irvine, California 92714 and publications annually. The IAFE is a professional non-profit Tel: (714) 752-8205 or financial association with members (800) 321-6338 worldwide specializing in new financial product design, derivative Fax: (714) 752-7444 structures, risk management strategies, arbitrage techniques, Email: Meetingmgt@aol.com and application of computational Web: http://www.ieee.org/nnc/cifer97 techniques to finance. ---------------------------------------------------------------------------- Payman Arabshahi CIFEr'97 Organizational Chair Tel: (206) 644-8026 Dept. Electrical Eng./Box 352500 Fax: (206) 543-3842 University of Washington Seattle, WA 98195 Email: payman@ee.washington.edu ---------------------------------------------------------------------------- From abonews@playfair.Stanford.EDU Thu Sep 5 00:10:10 1996 Received: from lucy.cs.wisc.edu (lucy.cs.wisc.edu [128.105.2.11]) by sea.cs.wisc.edu (8.6.12/8.6.12) with ESMTP id AAA02875 for ; Thu, 5 Sep 1996 00:10:02 -0500 Received: from TELNET-1.SRV.CS.CMU.EDU (TELNET-1.SRV.CS.CMU.EDU [128.2.254.108]) by lucy.cs.wisc.edu (8.6.12/8.6.12) with SMTP id AAA07322 for ; Thu, 5 Sep 1996 00:09:59 -0500 Received: from TELNET-1.SRV.CS.CMU.EDU by telnet-1.srv.cs.CMU.EDU id aa05481; 4 Sep 96 22:09:32 EDT Received: from DST.BOLTZ.CS.CMU.EDU by TELNET-1.SRV.CS.CMU.EDU id aa05479; 4 Sep 96 21:52:44 EDT Received: from DST.BOLTZ.CS.CMU.EDU by DST.BOLTZ.CS.CMU.EDU id aa03943; 4 Sep 96 21:52:27 EDT Received: from EDRC.CMU.EDU by B.GP.CS.CMU.EDU id aa27763; 4 Sep 96 18:53:24 EDT Received: from Playfair.Stanford.EDU by EDRC.CMU.EDU id aa00773; 4 Sep 96 18:52:57 EDT Received: from tukey.Stanford.EDU (tukey.Stanford.EDU [36.137.0.70]) by playfair.Stanford.EDU (8.7.1/8.6.10) with ESMTP id PAA18516 for ; Wed, 4 Sep 1996 15:52:54 -0700 From: Art Owen News Received: (abonews@localhost) by tukey.Stanford.EDU (8.7.1/8.6.10) id PAA05650 for Connectionists@cs.cmu.edu; Wed, 4 Sep 1996 15:52:54 -0700 Date: Wed, 4 Sep 1996 15:52:54 -0700 Message-Id: <199609042252.PAA05650@tukey.Stanford.EDU> To: Connectionists@cs.cmu.edu Subject: TR Available: Computer Experiments (noise free prediction) Address: http://playfair.stanford.edu/reports/owen File: main.ps for uncompressed PostScript main.ps.Z for compressed PostScript Authors: J. Koehler and A. Owen The above article is a survey paper on methods for computer experiments. These are noise free, usually continuous valued prediction problems, sometimes motivated by optimization problems in computer aided design. Without noise, how does one estimate the uncertainty in a given answer? A Bayesian approach places a process prior on the underlying function. An emerging frequentist approach samples the input space at random and propagates the sampling error. -Art Owen art@playfair.stanford.edu Replies should be sent to art@playfair not abonews where they might get lost among mailing list mail. My co-author is Jim Koehler: EMAIL: jkoehler@carbon.cudenver.edu http://www-math.cudenver.edu/~jkoehler From rosen@dragon.cs.utsa.edu Thu Sep 5 07:06:47 1996 Received: from lucy.cs.wisc.edu (lucy.cs.wisc.edu [128.105.2.11]) by sea.cs.wisc.edu (8.6.12/8.6.12) with ESMTP id HAA06971 for ; Thu, 5 Sep 1996 07:06:40 -0500 Received: from TELNET-1.SRV.CS.CMU.EDU (TELNET-1.SRV.CS.CMU.EDU [128.2.254.108]) by lucy.cs.wisc.edu (8.6.12/8.6.12) with SMTP id HAA09776 for ; Thu, 5 Sep 1996 07:06:38 -0500 Received: from TELNET-1.SRV.CS.CMU.EDU by telnet-1.srv.cs.CMU.EDU id aa05551; 4 Sep 96 22:24:19 EDT Received: from DST.BOLTZ.CS.CMU.EDU by TELNET-1.SRV.CS.CMU.EDU id aa05549; 4 Sep 96 22:14:45 EDT Received: from DST.BOLTZ.CS.CMU.EDU by DST.BOLTZ.CS.CMU.EDU id aa03960; 4 Sep 96 22:14:33 EDT Received: from RI.CMU.EDU by B.GP.CS.CMU.EDU id aa28764; 4 Sep 96 20:27:07 EDT Received: from runner.jpl.utsa.edu by RI.CMU.EDU id aa25119; 4 Sep 96 20:26:53 EDT Received: from dragon.cs.utsa.edu by runner.utsa.edu (5.0/SMI-SVR4) id AA04619; Wed, 4 Sep 1996 19:26:55 -0500 Received: from tachy.cs.utsa.edu by dragon.cs.utsa.edu (5.x/SMI-SVR4) id AA05170; Wed, 4 Sep 1996 19:27:26 -0500 Received: by tachy.cs.utsa.edu (SMI-8.6/SMI-SVR4) id TAA02780; Wed, 4 Sep 1996 19:27:23 -0500 Date: Wed, 4 Sep 1996 19:27:23 -0500 From: Bruce Message-Id: <199609050027.TAA02780@tachy.cs.utsa.edu> To: Connectionists@cs.cmu.edu Subject: Re: Paper announcements Cc: rosen@DRAGON.ITC.CS.CMU.EDU X-Sun-Charset: US-ASCII content-length: 1709 *** Paper Announcements *** The following paper is now available from my research page: http://www.cs.utsa.edu/faculty/rosen/rosen.html Comments are welcomed. ----------------------------------------------------------------------- Ensemble Learning using Decorrelated Neural Networks Bruce E. Rosen ftp://ringer.cs.utsa.edu/pub/rosen/decorrelate.ps.Z We describe a decorrelation network training method for improving the quality of regression learning in ``ensemble'' neural networks that are composed of linear combinations of individual neural networks. In this method, individual networks are trained by backpropagation to not only reproduce a desired output, but also to have their errors be linearly decorrelated with the other networks. Outputs from the individual networks are then linearly combined to produce the output of the ensemble network. We demonstrate the performances of decorrelated network training on learning the ``3 Parity'' logic function, a noisy sine function, and a one dimensional nonlinear function, and compare the results with the ensemble networks composed of independently trained individual networks (without decorrelation training). Empirical results show that when individual networks are forced to be decorrelated with one another the resulting ensemble neural networks have lower mean squared errors than the ensemble networks having independently trained individual networks. This method is particularly applicable when there is insufficient data to train each individual network on disjoint subsets of training patterns. To appear in: Connection Science, Special issue on Combining Estimators. From aonishi@bpe.es.osaka-u.ac.jp Thu Sep 5 20:45:29 1996 Received: from lucy.cs.wisc.edu (lucy.cs.wisc.edu [128.105.2.11]) by sea.cs.wisc.edu (8.6.12/8.6.12) with ESMTP id UAA22051 for ; Thu, 5 Sep 1996 20:45:22 -0500 Received: from TELNET-1.SRV.CS.CMU.EDU (TELNET-1.SRV.CS.CMU.EDU [128.2.254.108]) by lucy.cs.wisc.edu (8.6.12/8.6.12) with SMTP id UAA22631 for ; Thu, 5 Sep 1996 20:45:18 -0500 Received: from TELNET-1.SRV.CS.CMU.EDU by telnet-1.srv.cs.CMU.EDU id aa07329; 5 Sep 96 18:46:04 EDT Received: from DST.BOLTZ.CS.CMU.EDU by TELNET-1.SRV.CS.CMU.EDU id aa07327; 5 Sep 96 18:31:19 EDT Received: from DST.BOLTZ.CS.CMU.EDU by DST.BOLTZ.CS.CMU.EDU id aa04881; 5 Sep 96 18:30:52 EDT Received: from RI.CMU.EDU by B.GP.CS.CMU.EDU id aa06353; 5 Sep 96 4:52:47 EDT Received: from fsunc.bpe.es.osaka-u.ac.jp by RI.CMU.EDU id aa27443; 5 Sep 96 4:52:34 EDT Received: (from aonishi@localhost) by fsunc.bpe.es.osaka-u.ac.jp (8.6.12+2.4W/3.3Wb94061420) id RAA13861; Thu, 5 Sep 1996 17:52:20 +0900 Date: Thu, 5 Sep 1996 17:52:20 +0900 From: Toru Aonishi Message-Id: <199609050852.RAA13861@fsunc.bpe.es.osaka-u.ac.jp> To: Connectionists@cs.cmu.edu Subject: Paper Announcements *** Paper Announcements *** The following two papers on analysis of the dynamic link architecture are now available from my FTP site. ftp://ftp.bpe.es.osaka-u.ac.jp/pub/FukushimaLab/Papers/aonishi Comments/suggestions welcome, -Toru Aonishi (aonishi@bpe.es.osaka-u.ac.jp) =========================================================================== A Phase Locking Theory of Matching between Rotated Images by a Dynamic Link Architecture (Submitted to Neural Computation) Toru AONISHI, Koji KURATA and Takeshi MITO Pattern recognition invariant to deformation or translation can be performed with the dynamic link architecture proposed by von der Malsburg. The dynamic link has been applied to some engineering examples efficiently, but has not yet been analyzed mathematically. We propose two models of the dynamic link architecture. Both models are mathematically tractable. The first model can perform matching between rotated images. The second model can also do that, and can additionally detect common parts in a template image and in a data image. To analyze these models mathematically, we reduce each model's equation to a phase equation, showing the mathematical principle behind the rotating invariant matching process. We also carry out computer simulations to verify the mathematical theories involved. FTP-host: ftp.bpe.es.osaka-u.ac.jp FTP-pathname: /pub/FukushimaLab/Papers/aonishi/rotation_dy.ps.gz URL: ftp://ftp.bpe.es.osaka-u.ac.jp/pub/FukushimaLab/Papers/aonishi/rotation_dy.ps.gz 30 pages; 238Kb compressed. =========================================================================== Deformation Theory of Dynamic Link Architecture (Submitted to Neural Computation) Toru AONISHI, Koji KURATA Dynamic link is a self-organizing topographic mapping formed between a template image and a data image. The mapping tends to be continuous, linking two points sharing similar local features, which as a result, can lead to its deformation to some degree. Analyzing this deformation mathematically, we reduce the model equation to a phase equation, which clarifies the principles of this deformation process, the relation between high-dimensional models and low-dimensional ones. It also elucidates the characteristics of the model in the context of standard regularization theory. FTP-host: ftp.bpe.es.osaka-u.ac.jp FTP-pathname: /pub/FukushimaLab/Papers/aonishi/deform_dy.ps.gz URL: ftp://ftp.bpe.es.osaka-u.ac.jp/pub/FukushimaLab/Papers/aonishi/deform_dy.ps.gz 15 pages; 112Kb compressed. =========================================================================== From ruppin@math.tau.ac.il Fri Sep 6 15:23:02 1996 Received: from lucy.cs.wisc.edu (lucy.cs.wisc.edu [128.105.2.11]) by sea.cs.wisc.edu (8.6.12/8.6.12) with ESMTP id PAA08086 for ; Fri, 6 Sep 1996 15:22:35 -0500 Received: from TELNET-1.SRV.CS.CMU.EDU (TELNET-1.SRV.CS.CMU.EDU [128.2.254.108]) by lucy.cs.wisc.edu (8.6.12/8.6.12) with SMTP id PAA04883 for ; Fri, 6 Sep 1996 15:22:28 -0500 Received: from TELNET-1.SRV.CS.CMU.EDU by telnet-1.srv.cs.CMU.EDU id aa09107; 6 Sep 96 13:35:37 EDT Received: from DST.BOLTZ.CS.CMU.EDU by TELNET-1.SRV.CS.CMU.EDU id aa09105; 6 Sep 96 13:12:16 EDT Received: from DST.BOLTZ.CS.CMU.EDU by DST.BOLTZ.CS.CMU.EDU id aa05669; 6 Sep 96 13:10:27 EDT Received: from CS.CMU.EDU by B.GP.CS.CMU.EDU id aa02416; 6 Sep 96 8:37:57 EDT Received: from taurus.math.tau.ac.il by CS.CMU.EDU id aa19633; 6 Sep 96 8:37:12 EDT Received: from gemini.math.tau.ac.il (ruppin@gemini.math.tau.ac.il [132.67.64.2]) by taurus.math.tau.ac.il (8.7.3/8.7.3) with ESMTP id PAA24539; Fri, 6 Sep 1996 15:31:27 +0300 (GMT+0300) From: Eytan Ruppin Received: (from ruppin@localhost) by gemini.math.tau.ac.il (8.7.3/8.7.3) id PAA14712; Fri, 6 Sep 1996 15:31:24 +0300 (GMT+0300) Date: Fri, 6 Sep 1996 15:31:24 +0300 (GMT+0300) Message-Id: <199609061231.PAA14712@gemini.math.tau.ac.il> To: Connectionists@cs.cmu.edu, SMBnet@fconvx.ncifcrf.gov, ai-cbr-request@mailbase.ac.uk, ai-medicine@SMI.Stanford.EDU, ai-medicine@vuse.vanderbilt.edu, cbr-med@cs.uchicago.edu, clinical-psychology@mailbase.ac.uk, clinical-psychophysiology@mailbase.ac.uk, clp@cis.ohio-state.edu, comp-neuro@smaug.bbb.caltech.edu, compumed@sjuvm.stjohns.edu, computers-and-psychology@mailbase.ac.uk, elpida@turing.cs.ucy.ac.cy, int-neuro@lsr.nei.nih.gov, kdd-request@gte.com, medphys@radonc.duke.edu, ml@ics.UCI.EDU, neuron@CATTELL.psych.upenn.edu, news.announce.conferences@math.tau.ac.il, psycho-pharm@netcom.com, rberndt@umabnet.ab.umd.edu, reggia@cs.umd.edu, ruppin@math.tau.ac.il Subject: CFP:-Modeling-Brain-Disorders CALL FOR SUBMISSIONS Special Issue of the Journal "Artificial Intelligence in Medicine" (Published by Elsevier) Theme: COMPUTATIONAL MODELING OF BRAIN DISORDERS Guest-Editors: Eytan Ruppin & James A. Reggia (Tel-Aviv University) (University of Maryland) BACKGROUND As computational methods for brain modeling have advanced during the last several years, there has been an increasing interest in adopting them to study brain disorders in neurology, neuropsychology, and psychiatry. Models of Alzheimer's disease, epilepsy, aphasia, dyslexia, Parkinson's disease, stroke and schizophrenia have been recently studied to obtain a better understanding of the underlying pathophysiological processes. While computer models have the disadvantage of simplifying the underlying neurobiology and the pathophysiology, they also have remarkable advantages: They permit precise and systematic control of the model variables, and an arbitrarily large number of ``subjects''. They are open to detailed inspection, in isolation, of the influence of various metabolic and neural variables on the disease progression, in the hope of gaining insight into why observed behaviors occur. Ultimately, one seeks a sufficiently powerful model that can be used to suggest new pharmacological interventions and rehabilitative actions. OBJECTIVE OF SPECIAL ISSUE The objective of this special issue on modeling brain disorders is to report on the recent studies in this field. The main goal is to increase the awareness of the AI medical community to this research, currently primarily performed by members of the neural networks and `connectionist' community. By bringing together a series of such brain disorders modeling papers, we strive to produce a contemporary overview of the kinds of problems and solutions that this growing research field has generated, and to point to future promising research directions. More specifically, papers are expected to cover one or more of the following topics: -- Specific neural models of brain disorders, expressing the link between their pathogenesis and clinical manifestations. -- Computational models of pathological alterations in basic neural, synaptic and metabolic processes, that may relate to the generation of brain disorders in a significant manner. -- Applications of neural networks that shed light on the pathogenic processes that underlie brain disorders, or explore their temporal evolution and clinical course. -- Methodological issues involved in constructing computational models of brain disorders; obtaining sufficient data, visualizing high-dimensional complex behavior, and testing and validating these models. -- Bridging the apparent gap between functional imaging investigations and current neural modeling studies, arising from their distinct spatio-temporal resolution. SCHEDULE All the submitted manuscripts will be subject to a rigorous review process. The special issue will include 5 papers of 15-20 pages each, plus an editorial. Manuscripts should be prepared in accordance with the journal "submission guidelines" which are available on request, and may also be retrieved from http://www.math.tau.ac.il/~ruppin. November 15, 1996 Submission of tentative title and abstract to declare intension to submit paper. This should be done electronically, to ruppin@math.tau.ac.il. March 15, 1997 Receipt of full papers. Three copies of a manuscript should be sent to: Eytan Ruppin Department of Computer Science School of Mathematics Tel-Aviv University Tel-Aviv, Israel, 69978. August 1, 1997 Notification of acceptance October 1, 1997 Receipt of final-version of manuscripts June 1998 Publication of AIM special issue