Received: from kl.sri.com by RELAY.CS.NET id aa01291; 29 Jan 88 3:39 EST Date: Fri 29 Jan 1988 00:04-PST From: AIList Moderator Kenneth Laws Reply-To: AIList@KL.SRI.COM Us-Mail: SRI Int., 333 Ravenswood Ave., Menlo Park, CA 94025 Phone: (415) 859-6467 Subject: AIList V6 #18 - Seminars, Conference To: AIList@KL.SRI.COM Status: RO AIList Digest Friday, 29 Jan 1988 Volume 6 : Issue 18 Today's Topics: Seminar - The Psychology of Everyday Things (SUNY) & Combining O-O and DB Programming Languages? (Unisys) & Learning Search Control Knowledge (AT&T) & Thinkertools (BBN) & BREAD, FRAPPE, and CAKE: Automated Deduction (SRI), Conference - AAAIC88 Aerospace Applications of AI ---------------------------------------------------------------------- Date: Mon, 25 Jan 88 08:53:54 EST From: rapaport@cs.Buffalo.EDU (William J. Rapaport) Subject: Seminar - The Psychology of Everyday Things (SUNY) STATE UNIVERSITY OF NEW YORK AT BUFFALO The Steering Committee of the GRADUATE STUDIES AND RESEARCH INITIATIVE IN COGNITIVE AND LINGUISTIC SCIENCES PRESENTS DONALD A. NORMAN Institute for Cognitive Science University of California, San Diego THE PSYCHOLOGY OF EVERYDAY THINGS How do we manage the tasks of everyday life? The traditional answer is that we engage in problem solving, planning, and thought. How do we know what to do? Again, the traditional answer is that we learn, in part through experience, in part through instruction. I suggest that this view is misleading. Less planning and problem solving is required than is commonly supposed. Many tasks need never be learned: the proper behavior is obvious from the start. The problem space for most everyday tasks is shallow or narrow, not wide and deep as the tradi- tional approach suggests. The minimization of the problem space occurs because natural and contrived properties of the environment combine to constrain the set of possible actions. The effect is as if one had put the knowledge required to do a thing on the thing itself: the knowledge is in the world. I show that seven stages are relevant to the performance of an action, including three stages for execution of an act, three for evaluation, and a goal stage. Consideration of the rule of each stage, along with the principles of natural mappings and natural constraints, leads to a set of psychological principles for design. Couple these principles with the suggestion that most real tasks are shallow or narrow, and we start to have a psychology of everyday things and everyday actions. The talk itself is meant to be light and enjoyable. However, there are profound implications for the type of theory one develops for simulating cognitive computation. There are serious implications for massively parallel structures (what we call Parallel Distributed Processing or connectionist approaches), for memory storage and retrieval via descrip- tions or coarse coding, and, in general, for a central role for pattern matching, constraint satisficing, and nonsymbolic processing mechanisms in human cognition. But the main implications of the work are for the design of understandable and usable objects. Monday, February 1, 1988 4:00 P.M. Park 280, Amherst Campus There will also be an informal evening discussion that evening at David Zubin's house, 157 Highland St., at 8:00 P.M. Call Bill Rapaport (Com- puter Science, (716) 636-3193, 3180) for further information. ------------------------------ Date: Tue, 26 Jan 88 11:44:22 EST From: finin@PRC.Unisys.COM (Tim Finin) Subject: Seminar - Combining O-O and DB Programming Languages? (Unisys) AI Seminar UNISYS Knowledge Systems Paoli Research Center Paoli PA Can We Combine Object-Oriented and Database Programming Languages? Peter Buneman Computer and Information Science University of Pennsylvania The inadequate expressive power of the relational data model for many database representation tasks -- especially those that do not conform to requirements of traditional data processing -- has led several database systems developers to adopt an alternative "object-oriented" approach to the representation of data. But if we do this, must we necessarily sacrifice the the high-level languages and the considerable implementation technology that have been developed for relational databases? I shall argue that if we take a more liberal attitude to what a relation is, we can generalize relational languages, and even some of the ideas in relational database design, to work for sets of objects. A closely related problem is how we represent sets of objects as typed values in a programming language. If we can find such a representation, can data types be checked statically as in languages like Pascal and Ada, or must we live with the difficulties and dangers of run-time type checking? Some recent results by Atsushi Ohori indicate that it is not only possible to do static type checking, but that the types can be automatically inferred: the programmer does not even have to declare the data types! 2:00 pm Wednesday, February 3, 1988 BIC Conference Room Unisys Paloi Research Center Route 252 and Central Ave. Paoli PA 19311 -- non-Unisys visitors who are interested in attending should -- -- send email to finin@prc.unisys.com or call 215-648-7446 -- ------------------------------ Date: Wed, 27 Jan 14:47:03 1988 From: dlm%research.att.com@RELAY.CS.NET Subject: Seminar - Learning Search Control Knowledge (AT&T) Learning Effective Search Control Knowledge: An Explanation-Based Approach Steven Minton Carnegie-Mellon University Monday, February 1, 1988 10:30 AM AT&T Bell Laboratories - Murray Hill 3C-436 In order to solve problems more effectively with accumulating experience, a problem solver must be able to learn and exploit search control knowledge. In this talk, I will discuss the application of explanation-based learning (EBL) techniques for acquiring domain-specific control knowledge. Although previous research has demonstrated that EBL is a viable approach for acquiring control knowledge, in practice EBL may not always generate useful control knowledge. For control knowledge to be effective, the cumulative benefits of applying the knowledge must outweigh the cumulative costs of testing whether the knowledge is applicable. Generating effective control knowledge may be difficult, as evidenced by the complexities often encountered by human knowledge engineers. In general, control knowledge cannot be indiscriminately added to a system; its costs and benefits must be carefully taken into account. To produce effective control knowledge, an explanation-based learner must generate "good" explanations -- explanations that can be profitably employed to control problem solving. In this talk, I will discuss the utility of EBL and describe the PRODIGY system, a problem solver that learns by searching for good explanations. I will also briefly describe a formal model of EBL and a proof that PRODIGY's generalization algorithm is correct. Sponsor: Ron Brachman ------------------------------ Date: Thu 28 Jan 88 14:20:59-EST From: Marc Vilain Subject: Seminar - Thinkertools (BBN) BBN Science Development Program AI/Education Seminar Series Lecture THE THINKERTOOLS PROJECT: CAUSAL MODELS, CONCEPTUAL CHANGE, AND SCIENCE EDUCATION Barbara Y. White and Paul Horwitz BBN Labs, Education Dept. (BYWHITE@G.BBN.COM, PHORWITZ@G.BBN.COM) BBN Labs 10 Moulton Street 2nd floor large conference room 10:30 am, Thursday February 4 (NOTE UNUSUAL DAY) This talk will describe an approach to science education that enables sixth graders to learn principles underlying Newtonian mechanics, and to apply them in unfamiliar problem solving contexts. The students' learning is centered around problem solving and experimentation within a set of computer microworlds (i.e., interactive simulations). The objective is for students to acquire gradually an increasingly sophisticated causal model for reasoning about how forces affect the motion of objects. To facilitate the evolution of such a mental model, the microworlds incorporate a variety of linked alternative representations for force and motion, and a set of game-like problem solving activities designed to focus the students' inductive learning processes. As part of the pedagogical approach, students formalize what they learn into a set of laws, and critically examine these laws, using criteria such as correctness, generality, and parsimony. They then go on to apply their laws to a variety of real world problems. The approach synthesizes the learning of the subject matter with learning about the nature of scientific knowledge -- what are scientific laws, how do they evolve, and why are they useful? Instructional trials found that the curriculum is equally effective for males and females, and for students of different ability levels. Further, sixth graders taught with this approach do better on classic force and motion problems than high school students taught using traditional methods. ------------------------------ Date: Thu, 28 Jan 88 12:23:33 PST From: Amy Lansky Subject: Seminar - BREAD, FRAPPE, and CAKE: Automated Deduction (SRI) BREAD, FRAPPE, AND CAKE: THE GOURMET'S GUIDE TO AUTOMATED DEDUCTION Yishai A. Feldman (YISHAI@AI.AI.MIT.EDU) AI Laboratory, MIT 11:00 AM, WEDNESDAY, February 3 SRI International, Building E, Room EJ228 Cake is the knowledge representation and reasoning system developed as part of the Programmer's Apprentice project. Cake can be thought of as an active database, which performs quick and shallow deduction automatically; it supports both forward-chaining and backward-chaining reasoning. The Cake system has a layered architecture: the kernel of the system, called Bread (for Basic REAsoning Device), is a truth-maintenance system with equality and demons. Built on top of this is Frappe (for FRAmes in a ProPositional Engine), which implements a typed logic with special-purpose decision procedures for various algebraic properties of operators (such as commutativity and associativity), sets, partial functions, and structured objects (frames). Only the topmost layer of Cake, which implements the Plan Calculus, is specific to reasoning about programs. This talk will describe the architecture and features of Bread, Frappe, and Cake, including a transcript of a demonstration session. This is joint work with Charles Rich. VISITORS: Please arrive 5 minutes early so that you can be escorted up from the E-building receptionist's desk. Thanks! ------------------------------ Date: 26 Jan 88 16:54:00 EST From: "ETD2::WILSONJ" Reply-to: "ETD2::WILSONJ" Subject: Conference - AAAIC88 Aerospace Applications of AI AAAIC88 CALL FOR PAPERS AEROSPACE APPLICATIONS OF ARTIFICIAL INTELLIGENCE CONFERENCE 1988 With Neural Networks Aerospace Applications Special Interest Sessions Stouffer's Hotel, Dayton, OH, October 25-27, 1988 Particulars - Tutorials will be held on 24 Oct 88. Workshops will be held on 28 Oct 88. There will be exhibits by AI companies and related industries as well as product familiarization sessions. There will be up to 18 technical sessions in 5 half-day periods, luncheon speakers and a banquet. The 4th Aerospace Applications of Artificial Intelligence Conference will investigate a wide range of topics with heavy emphasis this year on neural network applications in aerospace. Topic areas for which timely, original, technical papers are solicited include: Integrating Neural Networks and Knowledge Processing with Neural Nets Expert Systems Robotics Neural Networks and Signal Processing Data Fusion/Sensor Fusion Machine Learning, Cognition & the Combinatorial Optimization for Cockpit Scheduling and Resource Control Machine Vision & Avionics Applications Natural Language Recognition and Neural Networks and Man-Machine Synthesis Interface Issues Self-Organization in Avionics Neural Network Development Tools Applied Adaptive-Resonance Applied Biological Models Cooperative and Competitive Network Parallel Processing & Neural Networks Dynamics in Aerospace Automatic Target Recognition Learning Theory and Techniques Back Propagation with Momentum, Simulation and Implementation of Shared Weights or Recurrent Neural Networks Network Architectures Technology - Microchips, Optics, etc. Expert System Development Tools Applications of Expert Systems in Aerospace Scheduling Manufacturing Operational and Maintenance Issues Design Automation Using Expert Systems Data Management Real Time Expert Systems Acquisition Management Knowledge Base Simulation Verification and Validation of ES Advanced Problem Solving Techniques Diagnostics and Fault Isolation ABSTRACT DEADLINE : 26 Feb 88 Authors are invited to submit abstracts of 500 words in any of the above topic areas. Please avoid acronyms or abbreviations in the title of the paper. A short biographical sketch of the author(s) to include citizenship, mailing address and telephone number must be included with the abstract. Final manuscripts for papers are due 19 Aug 88. James R. Johnson Send abstracts to: AFWAL/AAOR WPAFB, OH 45433 Sponsored by Dayton SIGART and the Association of Computing Machinery.