From in%@vtcs1 Sat Nov 15 01:28:09 1986 Date: Sat, 15 Nov 86 01:27:58 est From: vtcs1::in% To: ailist@sri-stripe Subject: AIList Digest V4 #258 Status: RO AIList Digest Wednesday, 12 Nov 1986 Volume 4 : Issue 258 Today's Topics: Query - Uncertainty and Belief in Expert Systems, Seminars - Is Probability Adequate? (MIT) & Uncertain Data Management (IBM) & Qualitative Reasoning about Mechanisms (SMU) & Diagnostic Systems (SMU) & Programming Descriptive Analogies by Example (Rutgers) & Explicit Contextual Knowledge in Learning (Rutgers) & Verification in Model-Based Recognition (MIT) & Formalizing the Notion of Context (SU), Conference - 2nd Knowledge Acquisition Workshop ---------------------------------------------------------------------- Date: 5 NOV 86 20:37-EST From: PJURKAT%SITVXA.BITNET@WISCVM.WISC.EDU Subject: SPRING RESEARCH SEMINAR AT STEVENS INSTITUTE OF TECHNOLOGY In association with two other faculty members of the Department of Management, I plan to offer a semester long research seminar, in the Spring 1987 semester, entitled REPRESENTATION OF UNCERTAINTY AND BELIEF IN EXPERT SYSTEMS To be covered are representations based on Bayesian theory, statistical inference and sampling distributions, discriminant functions, Schafer's theory of evidence, and fuzzy set theory. Participants will be asked to concentrate on finding and testing evidence which supports (or not) any of these theories as actually being related to the way experts deal with uncertainty and belief. The other faculty will review the representation work of cognitive science and experimental psychology. This note is to ask readers to pass on any recent work in these areas, particulary any experimental evidence on the actual workings of experts. We have the Kahneman, Slovic and Tversky book "Judgment under uncertainty: Heuristics and biases", published in 1982. I will post any interesting ideas and work that comes out of the seminar. Thank you for your consideration. Peter Jurkat (pjurkat@sitvxa) ------------------------------ Date: Wed 5 Nov 86 15:50:40-EST From: Rosemary B. Hegg Subject: Seminar - Is Probability Adequate? (MIT) UNCERTAINTY SEMINAR ON MONDAY Date: Monday, November 10, 1986 Time: 3.45 pm...Refreshments 4.00 pm...Lecture Place: NE43-512A UNCERTAINTY IN AI: IS PROBABILITY EPISTEMOLOGICALLY AND HEURISTICALLY ADEQUATE? MAX HENRION Carnegie Mellon New schemes for representing uncertainty continue to proliferate, and the debate about their relative merits seems to be heating up. I shall examine several criteria for comparing probabilistic representations to the alternatives. I shall argue that criticisms of the epistemological adequacy of probability have been misplaced. Indeed there are several important kinds of inference under uncertainty which are produced naturally from coherent probabilistic schemes, but are hard or impossible for alternatives. These include combining dependent evidence, integrating diagnostic and predictive reasoning, and "explaining away" symptoms. Encoding uncertain knowledge in predictive or causal form, as in Bayes' Networks, has important advantages over the currently more popular diagnostic rules, as used in Mycin-like systems, which confound knowledge about the domain and about inference methods. Suggestions that artificial systems should try to simulate human inference strategies, with all their documented biases and errors, seem ill-advised. There is increasing evidence that popular non-probabilistic schemes, including Mycin Certainty Factors and Fuzzy Set Theory, perform quite poorly under some circumstances. Even if one accepts the superiority of probability on epistemological grounds, the question of its heuristic adequacy remains. Recent work by Judea Pearl and myself uses stochastic simulation and probabilistic logic for propagating uncertainties through multiply connected Bayes' networks. This aims to produce probabilistic schemes that are both general and computationally tractable. HOST: PROF. PETER SZOLOVITS ------------------------------ Date: Thu, 06 Nov 86 10:19:35 PST From: IBM Almaden Research Center Calendar Subject: Seminar - Uncertain Data Management (IBM) IBM Almaden Research Center 650 Harry Road San Jose, CA 95120-6099 UNCERTAIN DATA MANAGEMENT L. A. Zadeh, Computer Science Division, University of California, Berkeley Computer Science Sem. Wed., Nov. 12 10:00 A.M. Room: Rear Audit. The issue of data uncertainty has not received much attention in the literature of database management even though the information resident in a database is frequently incomplete, imprecise or not totally reliable. Classical probability-based methods are of limited effectiveness in dealing with data uncertainty, largely because the needed joint probabilities are not known. Among the approaches which are more effective are (a) support logic programming which is Prolog-based, and (b) probabilistic logic. In our approach, uncertainty is modeled by (a) allowing the entries in a table to be set-values or, more generally, to be characterized as possibility distributions, and (b) interpreting a column as a source of evidence which may be fused with other columns. This model is closely related to the Dempster-Shafer theory of evidence and provides a conceptually simple method for dealing with some of the important types of uncertainty. In its full generality, the problem of uncertain data management is quite complex and far from solution at this juncture. Host: S. P. Ghosh ------------------------------ Date: WED, 10 oct 86 17:02:23 CDT From: leff%smu@csnet-relay Subject: Seminar - Qualitative Reasoning about Mechanisms (SMU) Dr. Benjamin Kuipers, Qualitative Reasoning About Mechanisms 10:00 AM, Friday, 7 November 1986 The first generation of diagnostic expert system is based on a simple model of knowledge: weighted links between observations and diagnoses. Experience with these systems has revealed a number of limitations in their performance due to the fact that they do not understand the mechanism by which a particular fault causes the associated observations. Recently developed methods for qualitative reasoning about these underlying mechanisms show promise of being able to extend the understanding, and hence the power, of diagnostic systems. The fundamental inference in qualitative reasoning derives the behavior of a mechanism from a description of its structure. Since both structure and behavior are represented in qualitative terms, this is essentially a qualitative abstraction of differential equations. I will derive in detail the QSIM approach to qualitative reasoning, and demonstrate a medical example in which QSIM predicts the behavior of a healthy mechanism, the "broken" mechanism corresponding to a particular disease, and the response of that broken mechanism to therapy. ------------------------------ Date: WED, 10 oct 86 17:02:23 CDT From: leff%smu@csnet-relay Subject: Seminar - Diagnostic Systems (SMU) Dr. William P. C. Ho Department of Computer Science and Engineering Southern Methodist University IEEE Computer Society Meeting, October 23, 1986 Diagnosis is the process of determining the cause (set of one or more physical component faults - "hypothesis" give the effect (set of one or more behavior deviations - "signature"), for a given mechanism. Ambiguity in interpreting fault signatures is the diagnosis problem. I am developing an approach for functional diagnosis of multiple component faults in mechanisms based on the "constraint satisfaction" paradigm (as opposed to "heuristic search" of "hypothesize and test"). Component faults and behavior deviations are both represented qualitatively by a set of 5 possible state values. Diagnostic reasoning is performed with these representations based on an effect calculus which combines more than one single fault effect into one single multiple fault effect quickly, without simulation. Diagnostic reasoning, encapsulated in a set of logical inference rules, is used to generate constraints, as implications of observed effects, which prune away subspaces of inconsistent hypotheses. The result is a complete set of consistent hypotheses which can explain all of the observed effects. ------------------------------ Date: 9 Nov 86 14:00:17 EST From: Tom Fawcett Subject: Seminar - Programming Descriptive Analogies by Example (Rutgers) On Tuesday November 25th, Henry Lieberman of MIT will speak on "Programming Descriptive Analogies by Example". The abstract follows. (The exact time will be decided later - it will probably be 10 AM in Hill-250.) Programming Descriptive Analogies By Example Henry Lieberman Artificial Intelligence Laboratory Massachusetts Institute of Technology This paper describes a system for "programming by analogy", called Likewise. Using this new approach to interactive knowledge acquisition, a programmer presents specific examples and points out which aspects of the examples are "slippable" to more general situations. The system constructs a general rule which can then be applied to "analogous" examples. Given a new example, the system can then construct an analogy with the old example by trying to instantiate new descriptions which correspond to the descriptions constructed for the first example. If a new example doesn't fit an old concept exactly, a concept can be generalized or specialized incrementally to make the analogy go through. Midway between "programming by example" and inductive inference programs, Likewise attacks the more modest goal of being able to communicate to the computer an analogy which is already understood by a person. Its operation on a typical concept learning task is presented in detail. ------------------------------ Date: 9 Nov 86 15:44:30 EST From: Smadar Subject: Seminar - Explicit Contextual Knowledge in Learning (Rutgers) Reminder: Dissertation Defense for Rich Keller Time and Place: Thursday, Nov. 13, 1:30 p.m., Hill 423 Committee: Tom Mitchell (chair) Thorne McCarty Lou Steinberg Jack Mostow Abstract: The Role of Explicit Contextual Knowledge in Learning Concepts to Improve Performance Richard M. Keller (KELLER@RED.RUTGERS.EDU) This dissertation addresses some of the difficulties encountered when using artificial intelligence-based, inductive concept learning methods to improve an existing system's performance. The underlying problem is that inductive methods are insensitive to changes in the system being improved by learning. This insensitivity is due to the manner in which contextual knowledge is represented in an inductive system. Contextual knowledge consists of knowledge about the context in which concept learning takes place, including knowledge about the desired form and content of concept descriptions to be learned (target concept knowledge), and knowledge about the system to be improved by learning and the type of improvement desired (performance system knowledge). A considerable amount of contextual knowledge is "compiled" by an inductive system's designers into its data structures and procedures. Unfortunately, in this compiled form, it is difficult for the learning system to modify its contextual knowledge to accommodate changes in the learning context over time. This research investigates the advantages of making contextual knowledge explicit in a concept learning system by representing that knowledge directly, in terms of express declarative structures. The thesis of this research is that aside from facilitating adaptation to change, explicit contextual knowledge is useful in addressing two additional problems with inductive systems. First, most inductive systems are unable to learn approximate concept descriptions, even when approximation is necessary or desirable to improve performance. Second, the capability of a learning system to generate its own concept learning tasks appears to be outside the scope of current inductive systems. To investigate the thesis, this study introduces an alternative concept learning framework -- the concept operationalization framework -- that requires various types of contextual knowledge as explicit inputs. To test this new framework, an existing inductive concept learning system (the LEX system [Mitchell et al. 81]) was rewritten as a concept operationalization system (the MetaLEX system). This dissertation describes the design of MetaLEX and reports results of several experiments performed to test the system. Results confirm the utility of explicit contextual knowledge, and suggest possible improvements in the representations and methods used by the system. ------------------------------ Date: Mon, 10 Nov 1986 21:08 EST From: JHC%OZ.AI.MIT.EDU@XX.LCS.MIT.EDU Subject: Seminar - Verification in Model-Based Recognition (MIT) THE USE OF VERIFICATION IN MODEL-BASED RECOGNITION David Clemens, MIT AI Lab The recognition of objects in images involves a gigantic and complex search through a library of models. Even for a single model, the correspondence between parts of the model and parts of the image can be difficult, especially if parts of the object may be occluded in the image. Verification is a general search strategy which can reduce the amount of processing required to find the best image/model match, but it cannot guarantee that the best match has been found. Verification is the Test phase of the familiar Hypothesize and Test paradigm, and is commonly used in the last stages of recognition to weed out final hypotheses. However, the concept can be applied more generally and used to drive the recognition process at much earlier stages. Also called "hypothesis-driven" recognition, this approach allows a more focused search for evidence to support, invalidate, or modify a hypothesis, thus decreasing the amount of data processed and improving the accuracy of the interpretation. Unfortunately, it requires a commitment to a finite set of initial hypotheses which must include an early version of correct hypotheses. Thus, there are trade-offs between hypothesis-driven modules and "data-driven" modules, which simply process all data uniformly without committing to early hypotheses. Several recognition systems will be discussed in this context, demonstrating the strengths and weaknesses of the two basic approaches applied to visual object recognition. Thursday, November 13, 4pm NE43 8th floor playroom ------------------------------ Date: 10 Nov 86 1108 PST From: Vladimir Lifschitz Subject: Seminar - Formalizing the Notion of Context (SU) Commonsense and Non-Monotonic Reasoning Seminar FORMALIZING THE NOTION OF CONTEXT John McCarthy Thursday, November 13, 4pm MJH 252 Getting a general database of common sense knowledge and expressing it in logic requires formalizing the notion of context. Since no context is absolutely general, any context must be elaboration tolerant and we discuss this notion. Another formalism that seems useful involves entering and leaving contexts; this is a generalization of natural deduction. ------------------------------ Date: Mon, 10 Nov 86 10:57:39 pst From: bcsaic!john@june.cs.washington.edu Subject: Conference - 2nd Knowledge Acquisition Workshop Call for Participation: 2ND KNOWLEDGE ACQUISITION FOR KNOWLEDGE-BASED SYSTEMS WORKSHOP Sponsored by the: AMERICAN ASSOCIATION FOR ARTIFICIAL INTELLIGENCE (AAAI) Banff, Canada October 19-23, 1987 A problem in the process of building knowledge-based systems is acquiring appropriate problem solving knowledge. The objective of this workshop is to assemble theoreticians and practitioners of AI who recognize the need for developing systems that assist the knowledge acquisition process. To encourage vigorous interaction and exchange of ideas the workshop will be kept small - about 40 participants. There will be individual presentations and ample time for technical discussions. An attempt will be made to define the state-of-the-art and the future research needs. Attendance will be limited to those presenting their work, one author per paper. Papers are invited for consideration in all aspects of knowledge acquisition for knowledge-based systems, including (but not restricted to) o Transfer of expertise - systems that obtain knowledge from experts. o Transfer of expertise - manual knowledge acquisition methods and techniques. o Apprenticeship learning systems. o Issues in cognition and expertise that affect the knowledge acquisition process. o Induction of knowledge from examples. o Knowledge acquisition methodology and training. Five copies of an abstract (up to 8 pages) or a full-length paper (up to 20 pages) should be sent to John Boose before April 15, 1987. Acceptance notices will be mailed by June 15. Full papers (20 pages) should be returned to the chairman by September 15, 1987, so that they may be bound together for distribution at the workshop. Ideal abstracts and papers will make pragmatic or theoretical contributions supported by a computer implementation, and explain them clearly in the context of existing knowledge acquisition literature. Variations will be considered if they make a clear contribution to the field (for example, comparative analyses, major implementations or extensions, or other analyses of existing techniques). Workshop Co-chairmen: Send papers via US mail to: John Boose Brian Gaines Advanced Technology Center Department of Computer Science Boeing Computer Services University of Calgary PO Box 24346 2500 University Dr. NW Seattle, Washington, USA 98124 Calgary, Alberta, Canada T2N 1N4 Send papers via express mail to: John Boose Advanced Technology Center Boeing Computer Services, Bldg. 33.07 2760 160th Ave. SE Bellevue, Washington, USA 98008 Program and Local Arrangements Committee: Jeffrey Bradshaw, Boeing Computer Services B. Chandrasekaran, Ohio State University Catherine Kitto, Boeing Computer Services Sandra Marcus, Boeing Computer Services John McDermott, Carnegie-Mellon University Ryszard Michalski, University of Illinois Mildred Shaw, University of Calgary ------------------------------ End of AIList Digest ******************** From in%@vtcs1 Sat Nov 15 01:28:23 1986 Date: Sat, 15 Nov 86 01:28:11 est From: vtcs1::in% To: ailist@sri-stripe Subject: AIList Digest V4 #259 Status: RO AIList Digest Wednesday, 12 Nov 1986 Volume 4 : Issue 259 Today's Topics: Administrivia - Splitting the List, Literature - Sentient-Computer Novels, Query - Knowledge-Base Portability, Logic Programming - Non-Monotonic Reasoning and Truth Maintenance, Application - Robotic Snooker, AI Tools - Franz Object-Oriented Packages & TCP from Xerox to UNIX System V, Ethics - Mathematics and Humanity & Why Train Machines & AI and the Arms Race ---------------------------------------------------------------------- Date: Tue 11 Nov 86 09:18:24-PST From: Ken Laws Reply-to: AIList-Request@SRI-AI.ARPA Subject: Splitting the List I recently received this request from an AIList reader: If there is a way to -just- get the seminar announcements periodically distributed to AIList, then I would like to be placed in that category. If this is not possible, then I wish to be removed from AIList completely. I have previously suggested that seminar and conference notices should be split out as a separate list (at least as long as other traffic remains so high), but no one has stepped forward to do the remailing. I haven't the energy to maintain two distribution lists. Volunteers are welcome. I'm sure there is still plenty of interest in other list topics. The NL-KR@Rochester list is doing fine, forwarding a great many natural-language messages that would not have appeared in AIList. IRList%VPI.CSNet has likewise been successful with information-retrieval topics. AI-Ed@SUMEX is alive and well. So is the Prolog Digest, which predates AIList. One reason for splitting the AIList is to reduce Arpanet traffic, which has been rather high lately, and to reduce costs for those who have to pay for the transmissions. Another is to reduce the difficulty for the next AIList moderator if I have to drop out. The best reason, though, is to boost discussion of the topics that most interest you. -- Ken Laws ------------------------------ Date: 7 Nov 86 20:57:17 GMT From: gknight@ngp.utexas.edu (Gary Knight) Subject: Canonical list of sentient computer novels Clarification of earlier posting, which is repeated below: 1) No robot novels, please; just non-ambulatory computers; and 2) No short works, just novels. --- I am trying to compile a canonical list of SF *novels* dealing with (1) sentient computers, and (2) human mental access to computers or computer networks. Examples of the two categories (and my particular favorites as well) are: A) SENTIENT COMPUTERS The Adolescence of P-1, by Thomas J. Ryan Valentina: Soul in Sapphire, by Joseph H. Delaney and Marc Stiegler Cybernetic Samurai, by (I forget) Coils, by Roger Zelazny B) HUMAN ACCESS True Names, by Vernor Vinge Neuromancer and Count Zero, by William Gibson Please send your lists to me by e-mail and I'll compile and post the ultimate canonical version. -- Gary Knight, 3604 Pinnacle Road, Austin, TX 78746 (512/328-2480). Biopsychology Program, Univ. of Texas at Austin. "There is nothing better in life than to have a goal and be working toward it." -- Goethe. ------------------------------ Date: 11 Nov 86 00:38:00 GMT From: u1100a!toh@bellcore.com (Tom O. Huleatt) Subject: Request for knowledge base portability info [Sorry if you see this twice -- postnews gagged on comp.ai, so I resubmitted.] Does anyone out there have experience with (or knowledge of) Knowledge Base portability issues? We have been using a home-grown rule-based system, and we are concerned about protecting our knowledge engineering investment as we move to other (more versatile) expert system shells. (These new systems will probably be rule-based, too.) I only have experience with our current system, so I'm not sure how much work is required to port one of our knowledge bases. I'd also be interested to hear any tips about what we could be doing with our knowledge bases now to increase their portability. Please send me email with suggestions (or pointers to ref. material). Thank you, Tom Huleatt [bellcore, ihnp4, pyuxww, allegra]!u1100a!toh Bell Communications Research Piscataway, NJ 08854 (201) 699-4506 ------------------------------ Date: Mon, 10 Nov 86 18:41:39 PST From: Tom Dietterich Subject: Non-monotonic reasoning and truth maintenance systems "These systems don't usually have any deductive power at all, they are merely constraint satisfaction devices." --David Etherington I am confused by this last sentence. Isn't constraint satisfaction a kind of inference? deKleer's ATMS and McAllester's RUP handle large portions (maybe all?) of propositional logic. --Tom Dietterich Department of Computer Science Oregon State University Corvallis, OR 97331 tgd%oregon-state.csnet ------------------------------ Date: Mon, 10 Nov 86 09:36:46 GMT From: Tony Conway Subject: Robotic Snooker In article <861020-061334-1337@Xerox> MJackson.Wbst@XEROX.COM writes: > >Over the weekend I caught part of a brief report on this on Cable News >Headlines. They showed a large robot arm making a number of impressive >shots, and indicated that the software did shot selection as well. >Apparently this work was done somewhere in Great Britain. Can someone >provide more detail? > >Mark Think that work was probably project by Richard Gregory (Brain & Perception Laboratory, Medical School, University of Bristol, Bristol, England) in conjunction with people in School of Engineering, Information Technology Research Centre, University of Bristol. Not sure if it has been written up anywhere yet. Richard Gregory is also active in starting up an interactive science centre (Bristol Exploratory): loosely based on the San Francisco Exploratorium. Cheers - 'Tony Conway ( @ucl-cs.arpa:tc@vd.rl.ac.uk ) Informatics, SERC Rutherford Appleton Laboratory, Chilton, Didcot, Oxon. OX11 0QX, England. ------------------------------ Date: Mon, 10 Nov 86 13:03:12 PST From: franz!fray!cox@ucbarpa.Berkeley.EDU (Charles A. Cox) Subject: Franz Object-Oriented Packages > Date: Wed, 5 Nov 86 13:08:28 EST > From: weltyc%cieunix@CSV.RPI.EDU (Christopher A. Welty) > Subject: Looking for Franz OO packages > > I am looking for information on Object Oriented extensions to > Franz Lisp. I know that someone (U of Maryland?) came out with a flavors > package for Franz, if someone can point me in the right direction there > it would be appreciated, as well as any info on other packages... Franz Inc. has a symbolics-compatible flavors package included in its versions of Franz Lisp (after Opus 42.0). I don't know much about the U of Maryland's system, but I believe they ship an entire Franz Lisp system (Opus 38) which includes their flavors package. The contact used to be Liz Allen (liz@tove.umd.edu). Other extensions to UC Berkeley's Franz Lisp put in by Franz Inc. include a common lisp compatible package system, multiple value returns, keywords, and hash tables. ------------------------------ Date: Mon 10 Nov 86 14:57:20-PST From: Christopher Schmidt Subject: TCP from Xerox to UNIX System V The TCP/IP package for Interlisp-D works for the most part, but usually requires a bit of fiddling to make work with any particular partner. Telnet generally works quite well with almost any host. I've used it to talk to unix 4.2, 4.3, System V, TOPS-20, and LispM telnet servers. FTP is a bit tricker and I usually have to run with the FTPDEBUG window on to figure out what to do. Logical pathname transformations are sometimes non-obvious and not all servers support the same set of commands. Since you ask about System V, I'll note that I've tested FTP against our Silicon Graphics Iris (System V, Excellan ethernet board) and found it to work OK. I don't use any TCPFTP server regularly, so I'm not the ideal reviewer. For nitty-gritty workstation questions, I recommend querying one of the workstation mailing lists rather than AIList. Eg. Bug-1100@SUMEX-AIM.Stanford.edu for Xerox d-machines (of which I am the moderator) SLUG@R20.UTexas.edu for Symbolics machines, or WorkS@Rutgers for workstations without their own mailing lists. --Christopher ------------------------------ Date: Fri, 7 Nov 86 14:54:26 EST From: "Col. G. L. Sicherman" Subject: Re: Mathematics and humanity In <8611050753.AA24198@ucbvax.Berkeley.EDU>, WADLISP7@CARLETON.BITNET writes: > The inhumanity of *most* mathematics? I would think that from the rest of > your message, what you would really claim is the inhumanity of *all* > mathematics -- for *all* of mathematics is entirely deviod of the questions > of what is morally right or morally wrong, entirely missing all matters of > human relationships. Mathematical theorems start by listing the assumptions, > and then indicating how those assumptions imply a result. This is the specialized mathematician's view of mathematics. The point is obviously sound, because mathematicians study mathematics as a thing apart. On the other hand, the mathematics that a herdsman uses to count sheep be- longs to the herdsman's life. It's not formally axiomatized, but it is human, because it is bound up with the natural human activity of growing food. To reinforce the point, many unlettered herdsmen have special numbers that they use _only_ for counting sheep. One can feel that to use those numbers for counting other things would be to endow those things with an inappropriate character of sheepliness. Modern mathematics rests on ignoring such "human" distinctions. The equals sign is the sine qua non of abstract mathematics--but it does not exist in human lives. The cry of "art for art's sake" produced generations of starving artists. What can we foresee from "math for math's sake?" ------------------------------ Date: Thu, 6 Nov 86 21:35:59 EST From: "Col. G. L. Sicherman" Subject: Re: Why train machines In article <861027-093832-2927@Xerox>, Ghenis.pasa@XEROX.COM writes: > > Why do we record music instead of teaching everyone how to sing? To > preserve what we consider top performance and make it easily available > for others to enjoy, even if the performer himself cannot be present and > others are not inclined to or capable of duplicating his work, but > simply wish to benefit from it. While I appreciate the point, it raises more questions.... 1. Why do we preserve top "performance?" The process of recording music redefines it as something perfectly repeatable--an effect that began with the invention of musical notation; jazz, and afterwards composers like Cage, tried to overturn this definition. But "performance" is also a social phenomenon, as it distinguishes between producers and consumers. The consequence of specialization is to retard progress by leaving the production of music to relatively few people. 2. Has not recorded music become a separate medium in its own right? Even a "faithful" recording involves a lot of electronic klugery. Most popular recordings no longer sound like, or can be performed as, live music. The second point has implications for A.I.! If you had a robot slave, how would you treat it? What would you become? ------------------------------ Date: Sat, 8 Nov 1986 13:38 EST From: LIN@XX.LCS.MIT.EDU Subject: AI and the Arms Race [I posted a message from AILIST on ARMS-D, and got back this reply.] Date: Saturday, 8 November 1986 12:55-EST From: ihnp4!utzoo!henry at ucbvax.Berkeley.EDU To: Arms-Discussion Re: Professionals and Social Responsibility for the Arms Race > ... This year, Dr. Weizenbaum of MIT was the chosen speaker... > The important points of the second talk can be summarized as : > 1) not all problems can be reduced to computation, for > example how could you conceive of coding the human > emotion loneliness. I don't want to get into an argument about it, but it should be pointed out that this is debatable. Coding the emotion of loneliness is difficult to conceive of at least in part because we don't have a precise definition of what the "emotion of loneliness" is. Define it in terms of observable behavior, and the observable behavior can most certainly be coded. > 2) AI will never duplicate or replace human intelligence > since every organism is a function of its history. This just says that we can't exactly duplicate (say) human intelligence without duplicating the history as well. The impossibility of exact duplication has nothing to do with inability to duplicate the important characteristics. It's impossible to duplicate Dr. Weizenbaum too, but if he were to die, I presume MIT *would* replace him. I think Dr. W. is on very thin ice here. > 5) technical education that neglects language, culture, > and history, may need to be rethought. Just to play devil's advocate, it would also be worthwhile to rethink non-technical education that covers language, culture, and history while completely neglecting the technological basis of our civilization. > 8) every researcher should assess the possible end use of > their own research, and if they are not morally comfortable > with this end use, they should stop their research... > He specifically referred to research in machine vision, which he > felt would be used directly and immediately by the military for > improving their killing machines... I'm afraid this is muddy thinking again. *All* technology has military applications. Mass-production of penicillin, a development of massive humanitarian significance, came about because of massive military funding in World War II, funding justified by the tremendous military significance of effective antibiotics. (WW2 was the first major war in which casualties from disease were fewer in number than those from bullets etc.) It's hard to conceive of a field of research which doesn't have some kind of military application. Henry Spencer @ U of Toronto Zoology {allegra,ihnp4,decvax,pyramid}!utzoo!henry ------------------------------ End of AIList Digest ******************** From in%@vtcs1 Thu Nov 20 00:36:28 1986 Date: Thu, 20 Nov 86 00:36:19 est From: vtcs1::in% To: ailist@sri-stripe.arpa Subject: AIList Digest V4 #260 Status: R AIList Digest Wednesday, 19 Nov 1986 Volume 4 : Issue 260 Today's Topics: Seminars - A Robust Approach to Plan Recognition (CMU) & Object-Oriented DBMSs (UPenn) & The Capacity of Neural Networks (UPenn) & BoltzCONS: Recursive Objects in a Neural Network (CMU) & Insight in Human Problem Solving (CMU) & Analogical and Deductive Reasoning (UCB) & Planning and Plan Recognition in Office Systems (Rutgers) & Logic Programming and Circumscription (SU) ---------------------------------------------------------------------- Date: 11 Nov 86 17:51:22 EST From: Steven.Minton@k.cs.cmu.edu Subject: Seminar - A Robust Approach to Plan Recognition (CMU) This week's speaker is Craig Knoblock. Usual time and place, 3:15 in 7220. Title: A Robust Approach to Plan Recognition Abstract: Plan recognition is the process of inferring an agent's plans and goals from his actions. Most of the previous work on plan recognition has approached this problem by first hypothesizing a single goal and then attempting to match the actions with a plan for achieving that goal. Unfortuantely, there are some types of problems where focusing on a single hypothesis will mislead the system. I will present an architecture for plan recognition that does not require the system to choose a single goal, but allows several hypotheses to be considered simultaneously. This architecture uses an assumption-based truth maintenance system to maintain both the observed actions and the predictions about the agent's plans and goals. ------------------------------ Date: Thu, 13 Nov 86 00:16 EST From: Tim Finin Subject: Seminar - Object-Oriented DBMSs (UPenn) DBIG Meeting 10:30 Friday November 14th 554 Moore School University of Pennsylvania DEVELOPMENT OF AN OBJECT-ORIENTED DBMS David Maier Oregon Graduate Center and Servio Logic Development Corp GemStone is an object-oriented database server developed by Servio Logic that supports a model of objects similar to that of Smalltalk. GemStone provides complex objects with sharing and identity, specification of behavioral aspects of objects, and an extensible data model. Those features came with the choice of Smalltalk as a starting point for the data model and its programming language, OPAL. However, Smalltalk is a single-user, memory-based system, and requires significant modifications to provide a multi-user, disk-based system with support for associative queries objects of arbitrary size. This presentation begins with a summary of the requirements for a database system to support applications such as CAD, office automation and knowledge bases. I next introduce the Smalltalk language and its data model, showing how they satisfy some of the requirements, and indicating which remain to be satisfied. I will outline the approach Servio took on the remaining requirements, describing the techniques used for storage management, concurrency, recovery, name spaces and associative access, as time permits. ------------------------------ Date: Thu, 13 Nov 86 23:12 EST From: Tim Finin Subject: Seminar - The Capacity of Neural Networks (UPenn) CIS Colloquium University of Pennsylvania 3pm Tuesday November 18 216 Moore School THE CAPACITY OF NEURAL NETWORKS Santosh S. Venkatesh University of Pennsylvania Analogies with biological models of brain functioning have led to fruitful mathematical models of neural networks for information processing. Models of learning and associative recall based on such networks illustrate how powerful distributed computational properties become evident as collective consequence of the interaction of a large number of simple processing elements (the neurons). A particularly simple model of neural network comprised of densely interconnected McCulloch-Pitts neurons is utilized in this presentation to illustrate the capabilities of such structures. It is demonstrated that while these simple constructs form a complete base for Boolean functions, the most cost-efficient utilization of these networks lies in their subversion to a class of problems of high algorithmic complexity. Specializing to the particular case of associative memory, efficient algorithms are demonstrated for the storage of memories as stable entities, or gestalts, and their retrieval from any significant subpart. Formal estimates of the essential capacities of these schemes are shown. The ultimate capability of such structures, independent of algorithmic approaches, is characterized in a rigourous result. Extensions to more powerful computational neural network structures are indicated. ------------------------------ Date: 12 November 1986 1257-EST From: Masaru Tomita@A.CS.CMU.EDU Subject: Seminar - BoltzCONS: Recursive Objects in a Neural Network (CMU) Time: 3:30pm Place: WeH 5409 Date: 11/18, Tuesday BoltzCONS: Representing and Transforming Recursive Objects in a Neural Network David S. Touretzky, CMU CSD BoltzCONS is a neural network in which stacks and trees are implemented as distributed activity patterns. The name reflects the system's mixed representational levels: it is a Boltzmann Machine in which Lisp cons cell-like structures appear as an emergent property of a massively parallel distributed representation. The architecture employs three ideas from connectionist symbol processing -- coarse coded distributed memories, pullout networks, and variable binding spaces, that first appeared together in Touretzky and Hinton's neural network production system interpreter. The distributed memory is used to store triples of symbols that encode cons cells, the building blocks of linked lists. Stacks and trees can then be represented as list structures, and they can be manipulated via associative retrieval. BoltzCONS' ability to recognize shallow energy minima as failed retrievals makes it possible to traverse binary trees of unbounded depth nondestructively without using a control stack. Its two most significant features as a connectionist model are its ability to represent structured objects, and its generative capacity, which allows it to create new symbol structures on the fly. A toy application for BoltzCONS is the transformation of parse trees from active to passive voice. An attached neural network production system contains a set of rules for performing the transformation by issuing control signals to BoltzCONS and exchanging symbols with it. Working together, the two networks are able to cooperatively transform ``John kissed Mary'' into ``Mary was kissed by John.'' ------------------------------ Date: 14 Nov 86 10:16:55 EST From: Jeffrey.Bonar@isl1.ri.cmu.edu Subject: Seminar - Insight in Human Problem Solving (CMU) An Interdiciplinary Seminar of the Computer Science Department and the Learning Research and Development Center UNIVERSITY OF PITTSBURGH AN INFORMATION PROCESSING ARCHITECTURE TO EXPLAIN INSIGHT IN HUMAN PROBLEM SOLVING STELLAN OHLSSON 10:00 AM TO 11:00, FRIDAY, JANUARY 9TH, 1987 LRDC AUDITORIUM, SECOND FLOOR REFRESHMENTS FOLLOWING There are currently four models of symbolic computation which are in frequent use in Cognitive Science work: applicative programming, logic programming, rule-based programming, and object oriented (frame based) programming. Each of these exhibit some general properties of human information processing, but neglects others. For example, LISP contains a model for the hiearchical structure of action, which Production Systems do not. What is needed for the simulation of human cognition is a new architecture which exhibits all of the properties which we know are characteristic of human cognition, and which "has" them in a natural way. An attempt at defining such an architecture will be presented. It has grown within a specific simulation attempt, namely to understand formally what happens in so-called "Aha"-experiences, moments of insight during problem solving. A theory has been constructed which explains such events within the information processing theory of problem solving as heuristic search. The theory is then implemented within the architecture described. An example of a run of the system will be described. For more information, call Cathy Rupp 624-3950 ------------------------------ Date: Mon, 17 Nov 86 13:55:23 PST From: admin%cogsci.Berkeley.EDU@berkeley.edu (Cognitive Science Program) Subject: Seminar - Analogical and Deductive Reasoning (UCB) BERKELEY COGNITIVE SCIENCE PROGRAM Cognitive Science Seminar - IDS 237A Tuesday, November 25, 11:00 - 12:30 2515 Tolman Hall Discussion: 12:30 - 1:30 2515 Tolman Hall ``Analogical and Deductive Reasoning" Stuart Russell Computer Science UC Berkeley The first problem I will discuss is that of analogical reason- ing, the inference of further similarities from known similari- ties. Analogy has been widely advertised as a method for apply- ing past experience in new situations, but the traditional approach based on similarity metrics has proved difficult to operationalize. The reason for this seems to be that it neglects the importance of relevance between known and inferred similarities. The need for a logical semantics for relevance motivates the definition of determinations, first-order expres- sions capturing the idea of relevance between generalized pro- perties. Determinations are shown to justify analogical infer- ences and single-instance generalizations, and to express an apparently common form of knowledge hitherto neglected in knowledge-based systems. Essentially, the ability to acquire and use determinations increases the set of inferences a system can make from given data. When specific determinations are unavailable, a simple statistical argument can relate similar- ity to the probability that an analogical solution is correct, in a manner closely connected to Shepard's stimulus generaliza- tion results. The second problem, suggested by and subsuming the first, is to identify the ways in which existing knowledge can be used to help a system to learn from experience. I describe a simple method for enumerating the types of knowledge (of which determinations are but one) that contribute to learn- ing, so that the monolithic notion of confirmation can be teased apart. The results find strong echoes in Goodman's work on induction. The application of a logical, knowledge-based approach to the problems of analogy and induction indicates the need for a system to be able to detect as many forms of regu- larity as possible in order to maximize its inferential capa- bility. The possibility that important aspects of common sense are captured by complex, abstract regularities suggests further empirical research to identify this knowledge. ------------------------------ Date: 17 Nov 86 12:59:16 EST From: BORGIDA@RED.RUTGERS.EDU Subject: Seminar - Planning and Plan Recognition in Office Systems (Rutgers) Computer Science Department Colloquium Date: Thursday November 20 Speaker: Professor Bruce Croft Title: Planning and Plan Recognition in Office Systems Affiliation: Department of Computer and Information Science, University of Massachusetts, Amherst Time: 10:00 a.m. [NOTE UNUSUAL TIME!!!] Place: Hill 705 Note: Refreshments will be served at 9:50 a.m. The office environment provides an ideal testbed for systems that attempt to represent and support complex, semi-structured and cooperative activities. It is typical to find a variety of constraints at different levels of abstraction on activities, objects manipulated by activities, and people that carry out the activities. In this talk, we will discuss the use of planning and plan recognition techniques to support an intelligent interface for an office system. In particular, we emphasise the use of object-based models, and the relationship between planning and plan execution. The types of exceptions that can occur with underconstrained plans will be described and some suggestions made about techniques for handling them. ------------------------------ Date: 17 Nov 86 1037 PST From: Vladimir Lifschitz Subject: Seminar - Logic Programming and Circumscription (SU) Commonsense and Non-Monotonic Reasoning Seminar LOGIC PROGRAMMING AND CIRCUMSCRIPTION Vladimir Lifschitz Thursday, November 20, 4pm MJH 252 The talk will be based on my paper "On the declarative semantics of logic programs with negation". A few copies of the paper are available in my office, MJH 362. ABSTRACT. A logic program can be viewed as a predicate formula, and its declarative meaning can be defined by specifying a certain Herbrand model of that formula. For programs without negation, this model is defined either as the Herbrand model with the minimal set of positive ground atoms, or, equivalently, as the minimal fixed point of a certain operator associated with the formula (Van Emden and Kowalski). These solutions do not apply to general logic programs, because a program with negation may have many minimal Herbrand models, and the corresponding operator may have many minimal fixed points. Apt, Blair and Walker and, independently, Van Gelder, introduced a class of general logic programs which disallow certain combinations of recursion and negation, and showed how to use the fixed point approach to define a declarative semantics for such programs. Using the concept of circumscription, we extend the minimal model approach to stratified programs and show that it leads to the same semantics. ------------------------------ End of AIList Digest ******************** From in%@vtcs1 Thu Nov 20 00:36:14 1986 Date: Thu, 20 Nov 86 00:36:06 est From: vtcs1::in% To: ailist@sri-stripe.arpa Subject: AIList Digest V4 #261 Status: R AIList Digest Wednesday, 19 Nov 1986 Volume 4 : Issue 261 Today's Topics: Queries - PEARL AI Package & Contextual Knowledge and Multilayer Learning & Logic Programming in APL & Cornell Synthesizer/Generator, Literature - Books Available for Review, Logic Programming - Nonmonotonic Reasoning and Truth Maintenance Systems, Science Fiction - Sentient Computers, Education - Cognitive Science Degree Programs, Ethics - AI and the Arms Race ---------------------------------------------------------------------- Date: Thu, 13 Nov 86 14:25:45 est From: rochester!tropix!dls@seismo.CSS.GOV (David L. Snyder ) Reply-to: tropix!dls@seismo.CSS.GOV (David L. Snyder ) Subject: pearl AI package A few questions about pearl (Package for Efficient Access to Representations in Lisp): Can anyone tell me what, if any, activity is going on with pearl these days? (Is the pearl-bugs mailing list still active?) Has anyone used it for non-toy problems? Any chance it'll be ported into common lisp? Is there something better that superceeds it (and is in the public domain)? Thanks! P.S. Try tropix!dls@rochester as an arpa address if other alternatives fail. ------------------------------ Date: Wed, 12 Nov 86 13:50 ??? From: MUKHOP%RCSJJ%gmr.com@RELAY.CS.NET Subject: Contextual Knowledge and Multi-layer Learning I read with interest the abstract for Richard M. Keller's talk, "The Role of Explicit Contextual Knowledge in Learning Concepts to Improve Performance" (V4 #258), part of which is reproduced below: > This dissertation addresses some of the difficulties encountered > when using artificial intelligence-based, inductive concept learning > methods to improve an existing system's performance. The underlying > problem is that inductive methods are insensitive to changes in the > system being improved by learning. This insensitivity is due to the > manner in which contextual knowledge is represented in an inductive > system. Contextual knowledge consists of knowledge about the context > in which concept learning takes place, including knowledge about the > desired form and content of concept descriptions to be learned (target > concept knowledge), and knowledge about the system to be improved by > learning and the type of improvement desired (performance system > knowledge). > ... > To investigate the thesis, this study introduces an alternative > concept learning framework -- the concept operationalization framework > -- that requires various types of contextual knowledge as explicit > inputs. >... Isn't this described in the literature as a two-layer learning system (multi-layer in the general case) of which Samuel's checkers player is one of the earliest examples? What are the differences, if any? Uttam Mukhopadhyay GM Research Labs ------------------------------ Date: Mon, 17 Nov 86 14:45 EST From: McHale@RADC-MULTICS.ARPA Subject: Logic programming in APL A while ago I heard of a system (from John Hopkins, I think) that combined logic programming with APL called APLLog. I would appreciate any pointers anyone could give me concerning this language. (User comments, software availability, underlying hardware, point of contact, etc.) Michael L. Mc Hale RADC/COES Griffiss AFB, NY 13441-5700 arpa% McHale RADC-Multics ------------------------------ Date: Tue, 18 Nov 86 16:00:56 pst From: Neil O'Neill Subject: Cornell Synthesizer/Generator (Need help) Does anyone have experience in running the Cornell Synthesizer/Generator? We could use some help with it if you know how to use it. Neil J. O'Neill ARPA: oneill@lll-tis-b.ARPA UUCP: {ihnp4,dual,sun}!lll-lcc!styx!oneill ------------------------------ Date: Tue, 18 Nov 86 08:01:20 -0500 From: sriram@ATHENA.MIT.EDU Subject: Books available for review The following books are available for review for the International Journal of AI in Engineering. If you are interested in acquiring a copy (and review it too), send mail to sriram@athena.mit.edu with your US mailing address. Please note that I have only single copies and books will be handed out on a first come basis. Machine Interpretation of Line Drawing Kokichi Sugihara MIT Press Introduction to Robotics. Mechanics and control. J. J. Craig Addison-Wesley Parallel Distributed Processing Vol 1 J.L. McClelland, D.E. Runelhart and the PDP research group MIT Press Paralell Distributed Processing Vol 2 J.L. McClelland,D.E. Runelhart and the PDP research group MIT Press The acquisition of Syntactic Knowledge R.C. Berwick MIT Press Computational Model of Discounts Edited by M. Brady and R. Berwick MIT Press Artificial Inteligence.The very Idea. J. Haugeland MIT Press Systems that learn D.N. Osherson, M. Srob and S. Weinstein MIT Press Robot Motion. Planning and control. Edited by M. Brady, J.M. Hollerbach,T.KL. Johnson, T. Lozano-Pevec MIT Press The measurement of Visual Motion Gllen Catherine Hildreth MIT Press A Geometric Investigation of Reach J.V. Korein MIT Press Robot Hands and the Mechanics of Manipulation M.T. Mason, J.K. Salisbury Jr MIT Press Theory and Practice of Robots and Manipulations Edited by A. Morecks,G. Bianchi and K. Kedziar MIT Press Robot Manipulators.Mathematics, Programming and Control Richard P. Paul MIT Press Expert Systems: Techniques, Tools and Applications P. Khalr and D. Waterman Addison-Wesley ------------------------------ Date: Thu, 13 Nov 86 10:06:11 est From: Randy Goebel LPAIG Subject: Re: Non-monotonic reasoning and truth maintenance systems > "These systems don't usually have any deductive power at all, > they are merely constraint satisfaction devices." > --David Etherington > > I am confused by this last sentence. Isn't constraint satisfaction > a kind of inference? deKleer's ATMS and McAllester's RUP handle > large portions (maybe all?) of propositional logic. > > --Tom Dietterich If one views constraint satisfaction as incremental model elimination, then there is a kind of inference going on, e.g., the number of models for p(X) & q(X) is reduced by adding the new constraint r(X), to get p(X) & q(X) & r(X). One can further see constraint satisfaction as inference by looking at Prolog puzzle solutions, where a list of constraints is posed as a goal, and the resolution prover must find a satisfying substitution; there is search involved, but satisfying substitutions are consequences of the axioms. Perhaps the best intuition about ``truth maintenance''-like systems is that they provide what is necessary for efficiently locating derivation steps that relied on assumptions. It's probably natural that any actual implementation blurs the distinction between the derivation maintenance and retrieval subsystem, and the prover that actually applies the inference rules to build derivations. Randy Goebel ------------------------------ Date: 12 Nov 86 15:43:00 GMT From: husc6!necntc!mirror!gabriel!inmet!sebes@eddie.mit.edu Subject: Re: Canonical list of sentient computer In the "sentient computer" class, there is Frank Herbert's _Destination Void_, which I recall as being notable not only for being a pretty good novel, but also for not appearing at all ridiculous or dated even 20-30 years after writing. In fact, some of the ideas mentioned are more in style now than then, or even a few years ago. --John Sebes ------------------------------ Date: 12 Nov 86 16:03:00 GMT From: husc6!necntc!mirror!gabriel!inmet!sebes@eddie.mit.edu Subject: Re: choosing grad schools Stanford has a program along the lines of that described at UCSD. The participating departments are CS, linguistics, philosophy, and psychology. There is a list of courses offered in those departments that count toward a course requirement for a phd in 'X and Cognitive Science' (I am not sure that that is the wording, but it is the gist). In addition to whatever course work you need to do in your department, you must take some number of those approved courses, with a certain distribution between your dept and the other three. Depending on your dept and how much course work you need to do there, it could be quite an undertaking. Also, it is a relatively recent thing, and I not sure how many people are actually involved in it. I found out about it simply by calling one of the depts and asking if that had any cogsci organization. Stanford also has a well-funded research center, the Center for the Study of Language and Intelligence (or something similar that spells CSLI ("Cicely")). --John Sebes ------------------------------ Date: Sun, 16 Nov 86 23:22:58 PST From: talmy%cogsci.Berkeley.EDU@berkeley.edu (Len Talmy) Subject: Cognitive Science degree programs at UC Berkeley In response to Don Norman's call for information, no, UC Berkeley does not have any degree-granting program in Cognitive Science either at the undergraduate or at the graduate level. So far, the most a student has been able to do is to make use of the special institutional apparatus for setting up a personally tailored degree program. However, we are now actively working on setting up a degree program at the undergraduate level. Even such a modest goal should take from one to two years, after all the committees have been formed and have analyzed the proposal. It was felt that a graduate degree program ought to be established only after an undergraduate one was in place and after some demand for Cognitive Science Ph.D.'s had developed. But the "Doctorate in X and Cognitive Science" formula is an interesting intermediate possibility, and we'll look into it. Len Talmy (coordinator, cognitive science program) talmy@cogsci.berkeley.edu ------------------------------ Date: Tue, 18 Nov 86 12:19:58 est From: "B. Lindsay Patten" Reply-to: "B. Lindsay Patten" Subject: Re: AI and the Arms Race In article LIN@XX.LCS.MIT.EDU writes: >[I posted a message from AILIST on ARMS-D, and got back this reply.] >From: ihnp4!utzoo!henry at ucbvax.Berkeley.EDU >Re: Professionals and Social Responsibility for the Arms Race [some valid objections to arguments made by Dr. Weizenbaum on problems with AI] >> 8) every researcher should assess the possible end use of >> their own research, and if they are not morally comfortable >> with this end use, they should stop their research... ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ >> He specifically referred to research in machine vision, which he >> felt would be used directly and immediately by the military for >> improving their killing machines... > >I'm afraid this is muddy thinking again. *All* technology has military >applications. [examples of good things that came out of military research] >It's hard >to conceive of a field of research which doesn't have some kind of military >application. > > Henry Spencer @ U of Toronto Zoology > {allegra,ihnp4,decvax,pyramid}!utzoo!henry This is by far the most common objection I've heard since Dr. Weizenbaum's lecture and one which I think avoids the point. Read the first three lines of point 8 above. The real point Dr. Weizenbaum was trying to make (in my opinion) was that we should weigh the good and bad applications of our work and decide which outweighs the other. The examples that he gave were just areas in which he personally believed the bad applications outweighed the good. He was very explicit that he was just presenting HIS personal opinions on the merits of these applications. Basically he said that if you feel your work will do more harm than good you should find another area to work in. My objection to his talk is that he seemed to want to weigh entire applications against one another. It seems to me that we should be examining the relative impact of our research in the applications which we approve of and in those we object to. Lindsay Patten |Cognitive Engineering Group (519) 746-1299| |Pattern Analysis and Machine Intelligence Lab lindsay@watsup| |University of Waterloo {decvax|ihnp4}!watmath!watvlsi!watsup!lindsay| ------------------------------ End of AIList Digest ********************