From ericr@mech.gla.ac.uk Fri Sep 1 11:26:27 1995 Received: from lucy.cs.wisc.edu by sea.cs.wisc.edu; Fri, 1 Sep 95 11:26:24 -0500; AA15190 Message-Id: <9509011626.AA17377@lucy.cs.wisc.edu> Received: from TELNET-1.SRV.CS.CMU.EDU by lucy.cs.wisc.edu; Fri, 1 Sep 95 11:26:21 -0500 Received: from TELNET-1.SRV.CS.CMU.EDU by telnet-1.srv.cs.CMU.EDU id aa00388; 1 Sep 95 5:19:32 EDT Received: from DST.BOLTZ.CS.CMU.EDU by TELNET-1.SRV.CS.CMU.EDU id aa00386; 1 Sep 95 5:11:49 EDT Received: from DST.BOLTZ.CS.CMU.EDU by DST.BOLTZ.CS.CMU.EDU id aa16137; 1 Sep 95 5:11:34 EDT Return-Path: Date: Fri, 1 Sep 1995 09:49:48 +0100 From: Eric Ronco To: Connectionists@cs.cmu.edu Subject: A modular neural network tech rep X-Sun-Charset: US-ASCII Dear all, We completed recently a technical report about modular neural networks. This one is actualy in the web page of our Centre for System and Control in the University of Glasgow (see the abstract below): http://www.mech.gla.ac.uk/Control/reports.html Ronco, E. and Peter Gawthrop, 1995. Modular Neural Networks: a state of the art. Tech. rep. CSC-95026, Centre for Systems and Control, faculty of engineering, Glasgow University. Any comments about this work would be well come. Regards, Eric Ronco Abstract: Title: Modular Neural Networks: a state of the art Author: Eric Ronco and Peter Gawthrop Keywords: Neural networks; Modularity; Global computation; Local computation; Clustering; Function approximation The use of ``global neural networks'' (as the back propagation neural network) and ``clustering neural networks'' (as the radial basis function neural network) leads each other to different advantages and inconvenients. The combination of the desirable features ot those two neural ways of computation is achieved by the use of Modular Neural Networks (MNN). In addition, a considerable advantage can emerge from the use of such a MNN: an interpreatable and relevant neural representation about the plant's behaviour. This very desirable feature for function approximation and especially for control problems, is what lake other neural models. This feature is so important that we introduce it as a way to differenciate MNN between other local computation models. However, to enable a systematic use of MNN three steps have to be achieved. First of all, the task has to be decomposed into subtasks, then the neural modules have to be properly organised considering the subtasks and finally a way of communication inter-modules has to be integrated in the whole architecture. We achieved a study of the main modular applications according to those steps. This study leads to the main fact that a systematic use of MNN depends on the type of task considered. The clustering networks and especially the Local Model Networks can be seen as MNN in the frame of classification or recognition problems. The Euclidean distance criterion that they apply to cluster the input space leads to a relevant decomposition according to the properties of those tasks. But, it is irrelevant to apply such a criteria in case of function approximation problems. As spatial clustering seems to be the only existing decomposing method, therefore, an ``ad hoc'' decomposition and organisation of the architecture is achieved in case of function approximation. So, to improve the systematic use of MNN in the framework of function approximation it is now essential to conceive a method of relevant task decomposition. From zhuh@helios.aston.ac.uk Fri Sep 1 18:51:41 1995 Received: from lucy.cs.wisc.edu by sea.cs.wisc.edu; Fri, 1 Sep 95 18:51:36 -0500; AA21278 Received: from TELNET-1.SRV.CS.CMU.EDU by lucy.cs.wisc.edu; Fri, 1 Sep 95 18:51:34 -0500 Received: from TELNET-1.SRV.CS.CMU.EDU by telnet-1.srv.cs.CMU.EDU id aa00884; 1 Sep 95 14:46:50 EDT Received: from DST.BOLTZ.CS.CMU.EDU by TELNET-1.SRV.CS.CMU.EDU id aa00882; 1 Sep 95 14:38:03 EDT Received: from DST.BOLTZ.CS.CMU.EDU by DST.BOLTZ.CS.CMU.EDU id aa16564; 1 Sep 95 14:37:21 EDT Received: from EDRC.CMU.EDU by B.GP.CS.CMU.EDU id aa01248; 1 Sep 95 7:52:44 EDT Received: from hermes.aston.ac.uk by EDRC.CMU.EDU id aa11721; 1 Sep 95 7:51:52 EDT Received: from email.aston.ac.uk by hermes.aston.ac.uk with SMTP (PP); Fri, 1 Sep 1995 12:53:26 +0100 Received: from darwin by email.aston.ac.uk with SMTP (PP) id <09437-0@email.aston.ac.uk>; Fri, 1 Sep 1995 12:53:07 +0100 Date: Fri, 1 Sep 1995 12:47:23 +0000 Message-Id: <26706.9509011147@sun.aston.ac.uk> From: H ZHU To: Connectionists@cs.cmu.edu Subject: 3 TechReports on Measuring Generalisation ... Content-Length: 2511 Is there any well-defined meaning to statements like "Learning rule A is better than learning rule B"? The answer is yes, as long as three things are specified: the prior, which is the distribution of problems to be solved; the information divergence, which tells how different the estimated distribution is from the true distribution; and the model, which is the space of all the representable solutions. The following three Technical Reports develop the necessary theory to evaluate and compare any neural network learning rules and other statistical estimators. ftp://cs.aston.ac.uk/neural/zhuh/discrete.ps.Z ftp://cs.aston.ac.uk/neural/zhuh/continuous.ps.Z ftp://cs.aston.ac.uk/neural/zhuh/generalisation.ps.Z Bayesian Invariant Measurements of Generalisation for Discrete Distributions Bayesian Invariant Measurements of Generalisation for Continuous Distributions Information Geometric Measurements of Generalisation by Huaiyu Zhu and Richard Rohwer ABSTRACT Neural networks can be considered as statistical models, and learning rules as statistical estimators. They should be compared in the framework of Bayesian decision theory, with information divergence as the loss function. This ensures coherence (An estimator is optimal if and only if it gives optimal estimates for almost all the data) and invariance (the optimality condition does not depend on one-one transforms in the input, output and parameter spaces). The main result is that the ideal optimal estimator is given as an appropriate average over the posterior. The optimal estimator restricted to any particular model is given by an appropriate projection of the ideal optimal estimator onto the model. The ideal optimal estimator is a sufficient statistic so that all the practical learning rules are its functions. They are also its approximations if preserving information in the data is the sole utility. This new theory of statistical inference retains many of the desirable properties of the least mean squares theory for linear Gaussian models, yet is applicable to any statistical estimation problem, including all the neural network learning rules (deterministic and stochastic, supervised, reinforcement and unsupervised). Comments are welcome and very much appreciated! -- Dr. Huaiyu Zhu zhuh@aston.ac.uk Neural Computing Research Group Dept of Computer Sciences and Applied Mathematics Aston University, Birmingham B4 7ET, UK From arbib@pollux.usc.edu Fri Sep 1 18:51:42 1995 Received: from lucy.cs.wisc.edu by sea.cs.wisc.edu; Fri, 1 Sep 95 18:51:38 -0500; AA21283 Received: from TELNET-1.SRV.CS.CMU.EDU by lucy.cs.wisc.edu; Fri, 1 Sep 95 18:51:36 -0500 Received: from TELNET-1.SRV.CS.CMU.EDU by telnet-1.srv.cs.CMU.EDU id ab00884; 1 Sep 95 14:47:58 EDT Received: from DST.BOLTZ.CS.CMU.EDU by TELNET-1.SRV.CS.CMU.EDU id ab00882; 1 Sep 95 14:38:05 EDT Received: from DST.BOLTZ.CS.CMU.EDU by DST.BOLTZ.CS.CMU.EDU id aa16569; 1 Sep 95 14:37:42 EDT Received: from EDRC.CMU.EDU by B.GP.CS.CMU.EDU id aa05296; 1 Sep 95 12:08:57 EDT Received: from pollux.usc.edu by EDRC.CMU.EDU id aa12734; 1 Sep 95 12:08:20 EDT Received: (arbib@localhost) by pollux.usc.edu (8.6.12/8.6.4) id JAA18825; Fri, 1 Sep 1995 09:08:09 -0700 Date: Fri, 1 Sep 1995 09:08:09 -0700 From: "Michael A. Arbib" Message-Id: <199509011608.JAA18825@pollux.usc.edu> To: Connectionists@cs.cmu.edu Subject: Website for Brain Theory and NN Handbook Cc: ritter@mit.edu The Handbook of Brain Theory and Neural Networks now has a Home Page on the Web. The URL is: http://www-mitpress.mit.edu/mitp/recent- books/comp/handbook-brain-theo.html This includes the preface, the complete table of contents, instructions on "How to Use this Book", and a Contributor list providing address, email, and article titles for all 341 authors. The ISBN is 0-262-01148-4 ARBHH The price is $150.00US till September 30, 1995, and $175 thereafter. Orders can be sent to the MIT Press at mitpress-orders@mit.edu Further MIT Press material is available on line at http://www-mitpress.mit.edu Best wishes Michael Arbib From stiber@cs.ust.hk Sat Sep 2 05:22:47 1995 Received: from lucy.cs.wisc.edu by sea.cs.wisc.edu; Sat, 2 Sep 95 05:22:37 -0500; AA26366 Received: from TELNET-1.SRV.CS.CMU.EDU by lucy.cs.wisc.edu; Sat, 2 Sep 95 05:22:35 -0500 Received: from TELNET-1.SRV.CS.CMU.EDU by telnet-1.srv.cs.CMU.EDU id aa01891; 2 Sep 95 4:36:05 EDT Received: from DST.BOLTZ.CS.CMU.EDU by TELNET-1.SRV.CS.CMU.EDU id aa01889; 2 Sep 95 4:24:46 EDT Received: from DST.BOLTZ.CS.CMU.EDU by DST.BOLTZ.CS.CMU.EDU id aa17333; 2 Sep 95 4:24:35 EDT Received: from RI.CMU.EDU by B.GP.CS.CMU.EDU id aa23372; 2 Sep 95 4:11:34 EDT Received: from [143.89.40.46] by RI.CMU.EDU id aa11865; 2 Sep 95 4:11:16 EDT Received: from cssu28.cs.ust.hk (stiber@cssu28.cs.ust.hk [143.89.40.28]) by cssu46.cs.ust.hk (8.6.12/8.6.9) with ESMTP id QAA06351; Sat, 2 Sep 1995 16:05:47 +0800 Received: (from stiber@localhost) by cssu28.cs.ust.hk (8.6.12/8.6.9) id QAA12069; Sat, 2 Sep 1995 16:05:42 +0800 Date: Sat, 2 Sep 1995 16:05:42 +0800 Message-Id: <199509020805.QAA12069@cssu28.cs.ust.hk> From: "Dr. Michael Stiber" To: comp-neuro@smaug.bbb.caltech.edu, connectionists@cs.cmu.edu, neuron-request@cattell.psych.upenn.edu, nel@UCSD.EDU Subject: Introducing the NeuroGeek WWW page Yes, yet another WWW page for you to place on the list of pages to check out some day when you have the time. The NeuroGeek page is for everyone interested in how computers can be applied to solving problems (or creating new ones) in the field of Computational Neuroscience. This includes differential equation integration methods, data analysis tools, applications of parallel computers, methods for simplifying models, repositories for code and information, etc. If I had to choose a unifying theme, it would be a focus on computational METHODS, as opposed to particular preparations, models, etc. We hope that this page will grow into a large virtual index showing HOW people are doing what they do. So please feel free to send pointers to information it may be lacking. The URL is: http://www.cs.ust.hk/faculty/stiber/neurogeek.html -- Dr. Michael Stiber stiber@cs.ust.hk Department of Computer Science tel: +852 2358 6981 The Hong Kong University of Science & Technology fax: +852 2358 1477 Clear Water Bay, Kowloon, Hong Kong finger me at: cssu28.cs.ust.hk or http://www.cs.ust.hk/faculty/stiber/bio.html cszm06.cs.ust.hk From pazzani@super-pan.ICS.UCI.EDU Sat Sep 2 17:58:52 1995 Received: from lucy.cs.wisc.edu by sea.cs.wisc.edu; Sat, 2 Sep 95 17:58:48 -0500; AA29679 Received: from paris.ics.uci.edu by lucy.cs.wisc.edu; Sat, 2 Sep 95 17:58:41 -0500 Received: from super-pan.ics.uci.edu by paris.ics.uci.edu id aa01990; 2 Sep 95 13:44 PDT To: ML-LIST:; Subject: Machine Learning List: Vol. 7, No. 15 Reply-To: ml@ics.uci.edu Date: Sat, 02 Sep 1995 13:18:23 -0700 From: Michael Pazzani Message-Id: <9509021344.aa01990@paris.ics.uci.edu> Machine Learning List: Vol. 7, No. 15 Saturday, September 2, 1995 Contents: AAAI Spring Symposium: Machine Learning in Information Access Conference on Evolutionary Programming (EP96) Call For Papers spline/ai/met paper ancts Call-for-papers: 1996 AAAI Spring Symposium Request for references to multiagent learning Predicting Learning Curves (Ph.D. thesis, online) Book announcement: GOAL-DRIVEN LEARNING Extension of deadline for AI/MATH-96 The Machine Learning List is moderated. Contributions should be relevant to the scientific study of machine learning. Mail contributions to ml@ics.uci.edu. Mail requests to be added or deleted to ml-request@ics.uci.edu. Back issues may be FTP'd from ics.uci.edu in pub/ml-list/V/ or N.Z where X and N are the volume and number of the issue; ID: anonymous PASSWORD: URL- http://www.ics.uci.edu/AI/ML/Machine-Learning.html ---------------------------------------------------------------------- From: Marti Hearst Date: Fri, 18 Aug 1995 17:06:26 PDT Subject: AAAI Spring Symposium: Machine Learning in Information Access Machine Learning in Information Access AAAI Spring Symposium Stanford, March 25-27, 1996 Paper Submission Date: Oct 31, 1995 As the volume and importance of the information available on the Internet continues to increase, there is a growing interest in information access in all areas of computer science. There has been substantial recent work on the application of machine learning techniques (e.g., inductive learning, genetic algorithms, and neural networks) to information access problems. For example, machine learning has been used to improve weights on terms for relevance feedback, to learn rules for filtering netnews articles, for automatic identification of hypertext links, and for text topic identification. Thus far, though, there has been no professional gathering devoted to investigating the use of machine learning techniques to improve access to textual information. The goal of this symposium is to provide the much-needed opportunity to develop new ideas and a well-defined community in this growing field. For this symposium, authors are asked to submit papers concerning the use of machine learning to enable or improve users' access to online information. Machine Learning techniques are especially appropriate for, but not limited to, the following information access tasks: Text Categorization and Segmentation Routing/Filtering Relevance Feedback Clustering User Preferences/Usage Pattern Analysis Browsing Multi-Source integration Papers can describe the use of machine learning on one or more of these tasks, or the development of new machine-learning algorithms tailored to information access tasks, or a comparison of learning algorithms vs. no learning on a given task. Although we expect that all accepted papers will provide objective evaluations of the main contributions of the described work (the use of standard test collections is encouraged, where applicable), this is an area with a lot of room for new ideas, and we would like this symposium to serve as an appropriate medium for publication of such work. Authors should make certain that the learning techniques they describe address the special issues that are associated with problems in information access. For example, because there are so many words in a typical text collection, ML problems in IA often involve a very large number >(10^5 - 10^6) of features. Characteristics of the available relevance assessments also make standard methods of evaluation problematic. Submission Information The symposium will consist of invited talks, paper presentations, and discussion sessions. Researchers from the information retrieval community are especially encouraged to participate. Interested participants should submit a short paper (5-8 pages maximum) addressing one or more of the research issues described above. If submitting via email, send either a URL pointing to a postscript version of the paper or the postscript copy itself to: mlia@parc.xerox.com. Or, send 5 hard copies to Marti Hearst Xerox PARC 3333 Coyote Hill Rd. Palo Alto, CA 94304 phone: (415) 812-4742 fax: (415) 812-4374 For further information, a web page for this symposium is located at http://www.xerox.com/PARC/mlia/mlia.html Program Committee: Richard K. Belew, University of California, San Diego Marti A. Hearst, Xerox PARC (CoChair) Haym Hirsh, Rutgers University (CoChair) Tom Mitchell, Carnegie Mellon University ------------------------------ From: "Peter J. Angeline" Date: Wed, 23 Aug 1995 13:30:03 -0400 Subject: Conference on Evolutionary Programming (EP96) Call For Papers Just a reminder that the submission deadline for the Fifth Annual Conference on Evolutionary Programming is September 26, 1995 and coming up fast. The conference will be held in San Diego from Feb 29 to March 3. We are featuring a $500 Best Student Paper Award at the conference this year. More information about submissions can be found at the conference homwpage at http://www.aic.nrl.navy.mil/galist/EP96/ Check it out. Questions about the conference or requests for text versions of the CFP can be addressed directly to me. ------------------------------ From: Grace Wahba Date: Wed, 23 Aug 95 14:20:00 -0500 Subject: spline/ai/met paper ancts Announcing new/revised manuscripts available on the web .... http/www.stat.wisc.edu/~wahba click on `TRLIST' also available by ftp in files listed at the end of this msg Luo, Z. and Wahba, G. "Hybrid Adaptive Splines" TR 947, June 1995, submitted. --- Combines ideas from smoothing splines and MARS to obtain spatial adaptivity, numerical efficiency. Numerical comparisons to wavelet simulations in Donoho et al, JRSSB 57, 1995. Approach parallels Orr, Neural Computation 7,1995, which we just became aware of-similar results on a different class of examples --. Wahba, G., Wang, Y., Gu, C., Klein, R. and Klein, B. " Smoothing Spline ANOVA for Exponential Families, with Application to the Wisconsin Epidemiological Study of Diabetic Retinopathy." May 1995, to appear, Annals of Statistics. --Expanded and *slightly revised* version of TR 940, December 1994. This paper was the basis for the Neyman Lecture given at the Annual Meeting of the Institute of Mathematical Statistics at Chapel Hill 1994, delivered by the first author. Collects results from SS-ANOVA, algorithm for choosing multiple smoothing parameters for Bernoulli data, implementing confidence intervals. Applies results to the estimation of four- year risk of progression of diabetic retinopathy, using data from the Wisconsin Epidemiological Study of Diabetic Retinopathy. (data available in GRKPACK documentation above). ***A discussion of the backfitting algorithm and SS-ANOVA has been added***--. Xiang, D. and Wahba, G. " A Generalized Approximate Cross Validation for Smoothing Splines with Non-Gaussian Data." TR 930, September 1994, submitted. --Another look at choosing the regularization parameter for Bernoulli data. Parallels work of Moody, Liu in the NN literature. Wahba, G., Johnson, D. R., Gao, F. and Gong, J. " Adaptive tuning of numerical weather prediction models: Part I: randomized GCV and related methods in three and four dimensional data assimilation." TR 920, April 1994. *revised and shortened* version to appear, Monthly Weather Review --describes how to use the randomized trace method to implement GCV in very large problems in numerical weather prediction, approximate solution of pde's with discrete noisy observations. Also available by ftp in gzipped postscript in ftp.stat.wisc.edu/pub/wahba in the following files, respectively has.ps.gz exptl.ssanova.rev.ps.gz gacv.ps.gz tuning-nwp.rev.ps.gz ------------------------------ From: Sandip Sen Date: Thu, 24 Aug 95 16:02:57 CDT Subject: Call-for-papers: 1996 AAAI Spring Symposium AAAI-96 Spring Symposium Series March 25-27, 1996 Stanford University, California Adaptation, Co-evolution and Learning in Multiagent Systems =========================================================== Coordination of multiple agents is essential for the viability of systems in which these agents share resources. Most of the research in Distributed Artificial Intelligence have concentrated on developing coordination strategies off-line. These pre-fabricated strategies can quickly become inadequate if the system designer's world model is incomplete/incorrect or if the environment can change dynamically. Learning and adaptation are invaluable mechanisms by which agents can evolve coordination strategies that meet the demands of the environments and the requirements of individual agents. The goal of this symposium is to focus on research that will address unique requirements for agents learning and adapting to work with other agents. Recognizing the applicability and limitations of current machine learning research as applied to multiagent problems as well as developing new learning and adaptation mechanisms particularly targeted to these class of problems will be of particular relevance to this symposium. We would particularly welcome new insights into this class of problems from other related disciplines, and thus would like to emphasize the inter-disciplinary nature of the symposium. Among others, papers of the following kind are welcome: -- Benefits of adaptive/learning agents over agents with fixed behavior in multiagent problems. -- Characterization of methods in terms of modeling power, communication abilities, and knowledge requirement of individual agents. -- Developing learning and adaptation strategies for environments with cooperative agents, selfish agents, partially cooperative agents. -- Analyzing and constructing algorithms that guarantee convergence and stability of group behavior. -- Co-evolving multiple agents with similar/opposing interests. -- Inter-disciplinary research from fields like organizational theory, game theory, psychology, sociology, economics, etc. Submission information ###################### The symposium will consist of individual presentations, invited talks, break-out group discussions, panels, and video sessions. Participants interested in presenting their work should send an extended abstract (12 point font, 5 pages or less) describing work in progress or completed work. Other interested participants should send a one-page description of their research interests with a short list of relevant publications. We would like to encourage submissions for video presentations and for working systems that can be used for hands-on demonstration during the symposium. We will accept only e-mail submissions of postscript files. Submissions should be sent to sandip@kolkata.mcs.utulsa.edu. Further information on this symposium can be found on the WWW at http://euler.mcs.utulsa.edu/~sandip/ss.html. Submissions for the symposia are due on October 31, 1995. Notification of acceptance will be given by November 30, 1995. Material to be included in the working notes of the symposium must be received by January 19, 1996. Organizing Committee #################### Sandip Sen (Chair), University of Tulsa, sandip@kolkata.mcs.utulsa.edu Devika Subramanian, Cornell University, devika@CS.Cornell.EDU Jeff Rosenschein, The Hebrew University, jeff@CS.HUJI.AC.IL John J. Grefenstette, Naval Research Laboratory, gref@AIC.NRL.Navy.Mil Michael N. Huhns, University of South Carolina, huhns@ece.sc.edu Tad Hogg, Xerox PARC, hogg@parc.xerox.com ------------------------------ From: Sandip Sen Date: Thu, 24 Aug 95 16:03:52 CDT Subject: Request for references to multiagent learning We recently completed a very successful workshop on "Adaptation and Learning in Multiagent Systems" in conjunction with IJCAI-95. The papers presented in the workshop are going to be published by Springer-Verlag as part of their Lecture Notes in Computers Science series, and will be edited by Gerhard Weiss and myself. I was the chair of the above-mentioned workshop, and will also be chairing a 1996 AAAI Spring Symposium on "Adaptation, Co-evolution, and Learning in Multiagent Systems." Because of the growing interest in this area of study, and to foster further research, I am putting together a bibliography of research in adaptation, evolution, and learning in multiagent systems. If you have published in this area, or know of the work of anyone else, please send me a reference to the published work. I would also appreciate if you can e-mail me a postscript copy of any relevant published paper. I welcome pointers to techreports, thesis work, workshop, conference, journal papers, etc. Thank you for your cooperation. Sandip Sen Department of Mathematical & Computer Sciences, University of Tulsa, 600 South College Avenue, Tulsa, OK 74104-3189, USA. (918) 631-2985 http://euler.mcs.utulsa.edu/~sandip/sandip.html e-mail: sandip@kolkata.mcs.utulsa.edu ------------------------------ From: "Carl M. Kadie" Date: Mon, 28 Aug 1995 15:43:12 -0500 Subject: Predicting Learning Curves (Ph.D. thesis, online) I'm happy to announce that my Ph.D. thesis Seer: Maximum Likelihood Regression for Learning-Speed Curves is available on-line via anonymous ftp from "ftp.cs.uiuc.edu", directory "pub/TechReports", file "UIUCDCS-R-95-1874.ps.Z". The URL is "ftp://ftp.cs.uiuc.edu/pub/TechReports/UIUCDCS-R-95-1874.ps.Z". The abstract is enclosed. - Carl ========================= Seer: Maximum Likelihood Regression for Learning-Speed Curves Carl Myers Kadie, Ph.D. Department of Computer Science University of Illinois at Urbana-Champaign, 1995 David C. Wilkins, Advisor The research presented here focuses on modeling machine-learning performance. The thesis introduces Seer, a system that generates empirical observations of classification-learning performance and then uses those observations to create statistical models. The models can be used to predict the number of training examples needed to achieve a desired level and the maximum accuracy possible given an unlimited number of training examples. Seer advances the state of the art with 1) models that embody the best constraints for classification learning and most useful parameters, 2) algorithms that efficiently find maximum-likelihood models, and 3) a demonstration on real-world data from three domains of a practicable application of such modeling. The first part of the thesis gives an overview of the requirements for a good maximum-likelihood model of classification-learning performance. Next, reasonable design choices for such models are explored. Selection among such models is a task of nonlinear programming, but by exploiting appropriate problem constraints, the task is reduced to a nonlinear regression task that can be solved with an efficient iterative algorithm. The latter part of the thesis describes almost 100 experiments in the domains of soybean disease, heart disease, and audiological problems. The tests show that Seer is excellent at characterizing learning-performance and that it seems to be as good as possible at predicting learning performance. Finally, recommendations for choosing a regression model for a particular situation are made and directions for further research are identified. ------------------------------ From: Ashwin Ram Date: Wed, 30 Aug 1995 14:01:35 -0400 Subject: Book announcement: GOAL-DRIVEN LEARNING GOAL-DRIVEN LEARNING ____________________ Edited by ASHWIN RAM and DAVID LEAKE MIT Press/Bradford Books, August 1995. ISBN 0-262-18165-7, 572 pp.__ 60 illus., $45 hardcover (Ordering information below) For more information, see http://www.cc.gatech.edu/cogsci/gdl.html TABLE OF CONTENTS _________________ Foreword by Professor Tom Mitchell Editors' Preface (see below) [1] Learning, Goals, and Learning Goals (Ram, Leake) (All of the above items are available from http://www.cc.gatech.edu/cogsci/gdl.html) Part I: Current state of the field [2] Planning to Learn (Hunter) [3] Quantitative Results Concerning the Utility of Explanation-Based Learning (Minton) [4] The Use of Explicit Goals for Knowledge to Guide Inference and Learning (Ram, Hunter) [5] Deriving Categories to Achieve Goals (Barsalou) [6] Harpoons and Long Sticks: The Interaction of Theory and Similarity in Rule Induction (Wisniewski, Medin) [7] Introspective Reasoning using Meta-Explanations for Multistrategy Learning (Ram, Cox) [8] Goal-Directed Learning: A Decision-Theoretic Model for Deciding What to Learn Next (desJardins) [9] Goal-Based Explanation Evaluation (Leake) [10] Planning to Perceive (Pryor, Collins) [11] Learning and Planning in PRODIGY: Overview of an Integrated Architecture (Carbonell, Etzioni, Gil, Joseph, Knoblock, Minton, Veloso) [12] A Learning Model for the Selection of Problem Solving Strategies in Continuous Physical Systems (Xia, Yeung) [13] Explicitly Biased Generalization (Gordon, Perlis) [14] Three Levels of Goal Orientation in Learning (Ng, Bereiter) [15] Characterising the Application of Computer Simulations in Education: Instructional Criteria (van Berkum, Hijne, de Jong, van Joolingen, Njoo) Part II: Current research and recent directions [16] Goal-Driven Learning: Fundamental Issues and Symposium Report (Leake, Ram) [17] Storage Side Effects: Studying Processing to Understand Learning (Barsalou) [18] Goal-Driven Learning in Multistrategy Reasoning and Learning Systems (Ram, Cox, Narayanan) [19] Inference to the Best Plan: A Coherence Theory of Decision (Thagard, Millgram) [20] Towards Goal-Driven Integration of Explanation and Action (Leake) [21] Learning as Goal-Driven Inference (Michalski, Ram) EDITORS' PREFACE ________________ In cognitive science, artificial intelligence, psychology, and education, a growing body of research supports the view that the learning process is strongly influenced by the learner's goals. Several experimental studies have shown that people with different goals process information differently. Studies in educational contexts have shown that different types of goals influence learning in different ways, and have attempted to use this insight in the design of effective educational environments. The importance of learner goals is supported by computational machine learning models, which provide functional arguments for goal-based focusing of learner effort, and from psychological evidence for the importance of student goals in educational settings. Investigators in each of these areas have independently pursued the common issues of how learning goals arise, how they affect learner decisions of when and what to learn, and how they guide the learning process. The fundamental tenet of goal-driven learning is that learning is largely an active and strategic process in which the learner, human or machine, attempts to identify and satisfy its information needs in the context of its tasks and goals, its prior knowledge, its capabilities, and environmental opportunities for learning. It is increasingly evident that investigation of goal-driven learning can benefit from a multidisciplinary effort employing diverse perspectives on a common research agenda. To this point, however, research in goal-driven learning has largely been confined to isolated efforts, with little framework to connect related results and to aid in their analysis. The purpose of this book is to establish such a framework, to collect and solidify existing results on goal-driven learning, and to point the way for future investigations of goal-driven learning. The book begins with a discussion of fundamental questions for goal-driven learning: the motivations for adopting a goal-driven model of learning, the basic goal-driven learning framework, the specific issues raised by the framework that a theory of goal-driven learning must address, the types of goals that can influence learning, the types of influences those goals can have on learning, and the pragmatic implications of the goal-driven learning model (chapter 1). The remainder of the book is divided into two parts. The first is a collection of recent research papers that serve as case studies in goal-driven learning. Each paper addresses a piece of the goal-driven learning puzzle, reflecting a particular research perspective from one of the several disciplines that have been investigating this area in recent years. These works address issues such as the justification of goal-driven learning models through functional arguments about the role and utility of goals in learning (chapters 2__4), the justification of such models through cognitive results (chapters 5, 6, 14), goal-based processes for deciding what to learn (chapters 7, 8) and for guiding learning and the learning process (chapters 4, 7, 9__13), and pragmatic implications of goal-driven learning for design of instructional environments (chapters 14, 15). The second part of the book is based on the Symposium on Goal-Driven Learning organized by David Leake and Ashwin Ram at the Fourteenth Annual Conference of the Cognitive Science Society in Bloomington, Indiana, in 1992. It presents an overview of the workshop discussion and a collection of papers from the symposium panelists representing their individual perspectives on fundamental issues and their proposals for fruitful future directions in goal-driven learning research. The works in this volume reflect both the diversity of goal-driven learning research and the fundamental relationship of different approaches within the broader goal-driven learning framework. Together, they provide a comprehensive overview of recent research in goal-driven learning and illuminate on-going investigations and open issues to provide a foundation for future study of goal-driven learning. ORDERING INFORMATION ____________________ E-mail: mitpress-orders@mit.edu Toll Free: (800) 356-0343 Fax: (617) 625-6660 Mail: The MIT Press, 55 Hayward Street, Cambridge, MA 02142-1399 WWW: http://www-mitpress.mit.edu/order-info.html FOR MORE INFORMATION ____________________ E-mail: ashwin@cc.gatech.edu or leake@cs.indiana.edu WWW: http://www.cc.gatech.edu/cogsci/gdl.html ------------------------------ From: ugur halici Date: Thu, 31 Aug 1995 20:07:19 +0400 Subject: "Extension of deadline for AI/MATH-96" Due to many requests, the submission deadline for the Fourth International Symposium on Artificial Intelligence and Mathematics (AI/MATH-96, Jan 3-5, Fort Lauderdale, Florida) has been EXTENDED to Friday, September 15, 1995. All submissions must be RECEIVED by this date. Electronic submissions (postscript) are preferred. Email extended abstracts (up to 10 double-spaced pages) to: selman@research.att.com - Henry Kautz and Bart Selman, program co-chairs CALL FOR PAPERS Fourth International Symposium on --------------------------------------- ARTIFICIAL INTELLIGENCE AND MATHEMATICS --------------------------------------- January 3-5, 1996, Fort Lauderdale Marina Marriott, Florida General Chair: Martin Golumbic, Bar-Ilan University, Ramat Gan Conference Chair: Frederick Hoffman, Florida Atlantic University Program co-chairs: Henry Kautz and Bart Selman AT&T Bell Labs Publicity Chair: Ugur Halici, Middle East Technical University Program Committee: Fahiem Bacchus (Waterloo) * Rachel Ben-Eliyahu (Technion) Endre Boros (Rutgers) * Alan Bundy (Edinburgh) * Marco Cadoli (Rome) James Crawford (Oregon) * Ernest Davis (NYU) * Rina Dechter (Irvine) Boi Faltings (EPFL) * Melvin Fitting (CUNY) * Eugene Freuder (New Hampshire) Erol Gelenbe (Duke) * Matt Ginsberg (Oregon) * Georg Gottlob (Vienna) Russell Greiner (Siemens) * Adam Grove (NEC) * Joseph Halpern (Almaden) Leo Joskowicz (Yorktown Heights) * Helene Kirchner (Nancy) Daphne Koller (Berkeley) * Richard Korf (UCLA) * Gerhard Lakemeyer (Bonn) Jean-Louis Lassez (Yorktown Heights) * Maurizio Lenzerini (Rome) Hector Levesque (Toronto) * Alon Levy (Bell Labs) * Vladimir Lifschitz (Texas) Alan Mackworth (UBC) * Witkor Marek (Kentucky) * Steven Minton (NASA Ames) Pandurang Nayak (NASA Ames) * Bernhard Nebel (Ulm) * Anil Nerode (Cornell) Ilkka Niemela (Helsinki) * Christos Papadimitriou (UCSD) * Jan Plaza (Miami) Teodor Przymusinski (UC Riverside) * Mauricio Resende (AT&T) Stuart Russell (Berkeley) * Elisha Sacks (Purdue) * Robert Schapire (Bell Labs) Yoav Shoham (Stanford) * Mark Stickel (SRI) * Moshe Tennenholtz (Technion) Leslie Valiant (Harvard) * Moshe Vardi (Rice) * Pierre Wolper (Liege) APPROACH OF THE SYMPOSIUM The International Symposium on Artificial Intelligence and Mathematics is the fourth of a biennial series. Our goal is to foster interactions among mathematics, theoretical computer science, and artificial intelligence. The meeting includes paper presentation, invited speakers, and special topic sessions. Topic sessions in the past have covered computational learning theory, nonmonotonic reasoning, and computational complexity issues in AI. The editorial board of the Annals of Mathematics and Artificial Intelligence serves as the permanent organizing committee for the series. SUBMISSIONS DEADLINE: SEPTEMBER 15th, 1995 FORMAT: Extended abstracts (up to 10 double-spaced pages). TO: Email (postscript): selman@research.att.com Or, send five copies to Bart Selman AT&T Bell Laboratories, Room 2T-414 600 Mountain Avenue Murray Hill, NJ 07974 USA Authors will be notified of acceptance or rejection on October 14th, 1995. Authors will be invited to submit within one month after the Symposium a final full length version of their paper to be considered for inclusion in a thoroughly refereed volume of the series Annals of Mathematics and Artificial Intelligence, J.C. Baltzer Scientific Publishing Co. SPONSORS The Symposium is partially supported by the Annals of Math and AI, Florida Atlantic University, and the Florida- Israel Institute. Other support is pending. Partial travel subsidies may be available to junior researchers. HOTEL AND TRAVEL INFORMATION Rooms have been blocked at the Fort Lauderdale Marina Marriott that are available to the participants at a rate of $95 per night, single or double occupancy for the Symposium, a huge savings against the ''rack rate.'' They must be booked directly with the hotel: Fort Lauderdale Marina Marriott, 1881 Southeast 17th Street, Fort Lauderdale, FL 33316. Phone: (305) 463-4000. The applicants must mention the symposium, and should probably mention Florida Atlantic University. The hotel directorship would like the applicants to specify arrival and departure dates and estimated time of arrival, room preference (single or double/double, smoking or non-smoking), credit card type to be used for payment including number and expiration date (this, or a one-night deposit, is absolutely necessary to hold rooms past 6PM). Delta Air Lines, Inc., in cooperation with the Fourth International Symposium on Artificial Intelligence and Mathematics, is offering special rates to the meeting. These fares are based on Delta's published round-trip fares within the United States and Canada, San Juan, Nassau, Bermuda, St. Thomas, and St. Croix. To take advantage of these discounts, call your travel agent or call Delta at 1-800-241-6760, for reservations 7:30 a.m.-11:00 p.m., Mon-Fri, 8:30 a.m.-11:00 p.m., Sat/Sun Eastern time, and refer to file number XM0039. FURTHER INFORMATION AND FUTURE ANNOUNCEMENTS Contact: Frederick Hoffman, Florida Atlantic University, Department of Mathematics, PO Box 3091, Boca Raton, FL 33431, USA E-mail: hoffman@acc.fau.edu or hoffman@fauvax.bitnet ------------------------------ End of ML-LIST (Digest format) ****************************************