@SANDIA.ARPA@lanl-a.UUCP Stan Steinberg stanly@unmvax ******************************************************************* Rio Grande Chapter SIGNUM Meeting Year end meeting and election of officers Date: Tuesday, May 8, 1984 Speakers: Kathie Hiebert Dodd and Barry Marder - Sandia Applied AI - "Brave New World" or "Catastrophe Theory Revisited"? Barry Marder Last year an effort was initiated at Sandia to develop a core of expertise in the field of artificial intelligence. One area of investigation has been expert system technology, which has been largely responsible for the present explosive growth of interest in AI. An expert system is a program that catalogs and makes readily available expert knowledge in a field. Such a system has been built and implemented at Sandia to aid in the design of electrical cables and connectors. The speaker will describe this system and offer some observations on artificial intelligence in general. VEHICLE IDENTIFICATION -- A FRAME BASED SYSTEM Kathie Hiebert Dodd Software has been developed that, when given certain characteristics from a scene such as the location of wheels, can identify vehicles. The image processing, ie extracting the characteristics from the scene is still done primarily on a VAX. Given the features a frame based code using "flavors" in the Zetalisp language on a Symbolics 3600 does the vehicle identification. The main emphasis of the talk will be on the aspects of a frame based expert system, in particular the use of "flavors" and "deamons". Location: The Establishment - Albuquerque Dukes Sports Stadium Price: 10.50 per person - serving Prime Rib (I think) Social Hour : 5:30 P.M., Dinner: 6:00 P.M., Talks: 7:00 P.M. PLEASE LET JOHN WISNIEWSKI KNOW BY NOON MONDAY THE 7TH IF YOU ARE COMING TO DINNER. If no answer leave a message with EVA 844-7747. ------------------------------ Date: 4 May 1984 1316-EDT From: Geoff Hinton Subject: Seminar - Learning in Production Systems [Forwarded from the CMU-AI bboard by Laws@SRI-AI.] The AI seminar on May 8 will be given by John Holland of the University of Michigan. Title: Learning Algorithms for Production Systems Learning, broadly interpreted to include processes such as induction, offers attractive possibilities for increasing the flexibility of rule-based systems. However, this potential is likely to be realized only when the rule-based systems are designed ab initio with learning in mind. In particular, there are substantial advantages to be gained when the rules are organized in terms of building blocks suitable for manipulation by the learning algorithms (taking advantage of the principles expounded by Newell & Simon). This seminar will concentrate on: 1. Ways of inducing useful building blocks and rules from experience, and 2. Learning algorithms that can exploit these possibilities through "apportionment of credit" and "recombination" of building blocks. ------------------------------ Date: Sat 5 May 84 18:45:28-PDT From: Benjamin Grosof Subject: Seminars - Nonmonotonic Reasoning [Forwarded from the CSLI bboard by Laws@SRI-AI.] Our regular meeting time and place is Wednesdays 1-2pm (with some runover to be expected), in Redwood Hall Room G-19. [...] Wednesday, May 16: Drawing A Line Around Circumscription David Etherington University of British Columbia, Vancouver The Artificial Intelligence community has been very interested in the study of reasoning in situations where only incomplete information is available. Predicate Circumscription and Domain Circumscription provide tools for nonmonotonic reasoning in such situations. However, not all of the problems which might be expected to yield to circumscriptive inference are actually addressed by the techniques which have been developed thus far. We outline some unexpected areas where existing techniques are insufficient. Wednesday, May 23 DEFAULT REASONING AS CIRCUMSCRIPTION A Translation of Default Logic into Circumscription OR Maximizing Defaults Is Minimizing Predicates Benjamin Grosof of Stanford Much default reasoning can be formulated as circumscriptive. Using a revised version [McCarthy 84] of circumscription [McCarthy 80], we propose a translation scheme from default logic [Reiter 80] into circumscription. An arbitrary "normal" default theory is translated into a corresponding circumscription of a first-order theory. The method is extended to translating "seminormal" default theories effectively, but is less satisfactorily concise and elegant. Providing a translation of seminormal default logic into circumscription unifies two of the leading formal approaches to nonmonotonic reasoning, and enables an integration of their demonstrated applications. The naturalness of default logic provides a specification tool for representing default reasoning within the framework of circumscription. ------------------------------ Date: Fri, 4 May 84 15:52 PDT From: Brian Reid Subject: 12th POPL Call for Papers Call for Papers: 12th P