Date: Sun 17 Jan 1988 21:41-PST From: AIList Moderator Kenneth Laws Reply-To: AIList@SRI.COM Us-Mail: SRI Int., 333 Ravenswood Ave., Menlo Park, CA 94025 Phone: (415) 859-6467 Subject: AIList V6 #11 - Seminars To: AIList@SRI.COM Status: R AIList Digest Monday, 18 Jan 1988 Volume 6 : Issue 11 Today's Topics: Seminars - Rational Choice and Cognitive Illusion (HP) & Cooperative Inference Machine: CHI (SRI) & OBJ as a Theorem Prover (SRI) & Four-Valued Semantics for Terminological Logics (AT&T) ---------------------------------------------------------------------- Date: Mon 11 Jan 88 09:38:06-PST From: Oscar Firschein Subject: Seminar - Rational Choice and Cognitive Illusion (HP) HP Labs Colloquium: Prof. Amos Tversky Subject: Rational Choice and Cognitive Illusion The analysis of decision and judgment under uncertainty reveals pervasive and systematic departures from the rational theory of judgment and choice. In particular, people exhibit overconfidence in action and belief, susceptability to framing effects, and inconsistent attitudes toward risk. These pheenomena are traced to the operation of a limited number of heuristic principles, which are generally useful but often produce illusion and bias. Time: 4 P.M. Thursday, Feb. 4, 1988 Place: Hewlett-Packard 5M Auditorium 1501 Page Mill Road, Palo Alto Non-HP employees: Welcome! Please come to the lobby on time so that you may be escorted to the 5M auditorium Refreshment: A wine and cheese reception will follow. ------------------------------ Date: Thu, 14 Jan 88 10:05:24 PST From: seminars@csl.sri.com (contact lunt@csl.sri.com) Subject: Seminar - Cooperative Inference Machine: CHI (SRI) SRI COMPUTER SCIENCE LAB SEMINAR ANNOUNCEMENT: A Cooperative High-Performance Sequential Inference Machine: CHI Akihiko Konagaya Computer Systems Research Laboratory NEC Corporation, Japan Tuesday, January 26 at 11:00 am SRI International, Conference Room B, Building A This talk will describe the design principles and new compiler technique of CHI-II, a deskside backend inference machine, which has been developed as part of the Fifth Generation Computer System (FGCS) Project in Japan. CHI-II achieves 500 KLIPS for deterministic append program execution by means of specialized hardware and machine code optimization techniques. The architecture also executes interpretive predicates efficiently. The merit of the micro-programmable "Wide Spectrum Instruction Set," as opposed to the "Reduced Instruction Set (RISC)," in terms of logic programming language execution, will be addressed. One of the most important features of the CHI-II hardware is its large capacity main memory (600MBytes) for a single user. Cutting-edge memory device technology makes it possible to develop a large-capacity, but small-size, main memory system comparable to an ordinal secondary storage. This large-capacity memory allows us to store massive amounts of data, such as the knowledge base of an expert system, in memory, and greatly contributes to achieving high performance in large-scale application systems by eliminating the overhead of virtual memory management. The CHI software system aims at integrating heterogeneous programming environments on a workstation in a distributed operating system fashion. The software enables CHI to act like a stand-alone computer without input/output devices, rather than as an accelerator of a host machine. An object-oriented means of intermachine communication, called "virtual objects," greatly simplifies the realization of bus-transparent input/output operations, such as a remote file system and a remote window system. The multiple process facility, although rather conservative, enables us to develop practical system programs dedicated to logic programming. It is also expected to be used as a good vehicle in which to study parallel logic programming. NOTE FOR VISITORS TO SRI: Please arrive at least 10 minutes early in order to sign in and be shown to the conference room. SRI is located at 333 Ravenswood Avenue in Menlo Park. Visitors may park in the visitors lot in front of Building A (red brick building at 333 Ravenswood Ave) or in the conference parking area at the corner of Ravenswood and Middlefield. The seminar room is in Building A. Visitors should sign in at the reception desk in the Building A lobby. IMPORTANT: Visitors from Communist Bloc countries should make the necessary arrangements with Fran Leonard, SRI Security Office, (415) 859-4124, as soon as possible. ------------------------------ Date: Thu, 14 Jan 88 10:04:42 PST From: seminars@csl.sri.com (contact lunt@csl.sri.com) Subject: Seminar - OBJ as a Theorem Prover (SRI) SRI COMPUTER SCIENCE LAB SEMINAR ANNOUNCEMENT: OBJ AS A THEOREM PROVER Joseph A. Goguen Computer Science Laboratory SRI International Monday, January 25 at 4:00 pm SRI International, Conference Room B, Building A This talk has two goals: to introduce OBJ, and to present some techniques for using OBJ as a theorem prover. OBJ is a wide-spectrum, first order functional programming language rigorously based on *order-sorted* equational logic, which provides a notion of *subtype* to support overloading, coercion, multiple inheritance, and exception handling. This rigorous semantic basis allows both a declarative programming style, and the direct use of OBJ for theorem proving. Parameterized programming is a powerful technique for software design, production, reuse and maintenance, involving abstraction through two kinds of module: *objects* to encapsulate executable code, and in particular to define abstract data types; and *theories* to specify both syntactic and semantic structure of modules. Each kind of module can be parameterized, where actual parameters are modules. Modules can also import other modules, yielding hierarchies of parameterized modules. Interfaces of parameterized modules are defined by theories. For parameter instantiation, a *view* binds the formal entities in an interface theory to actual entities in a module, and also asserts satisfaction of the theory by the module. Views are first class citizens that can be named, can import modules, and can even be parameterized. *Module expressions* allow complex instantiations and may include commands that transform already defined modules. Typical higher order programming examples can be captured with just first order functions, by the systematic use of parameterized programming. Some examples are given, including a hardware verification example. New results include a simple but useful extension of first order equational logic to allow quantification over arbitrary function symbols, a perhaps surprising technique for proving such equations using only ground term reduction, and some general induction principles. All this provides a very powerful first order calculus for reasoning about (first order) functions. NOTE FOR VISITORS TO SRI: Please arrive at least 10 minutes early in order to sign in and be shown to the conference room. SRI is located at 333 Ravenswood Avenue in Menlo Park. Visitors may park in the visitors lot in front of Building A (red brick building at 333 Ravenswood Ave) or in the conference parking area at the corner of Ravenswood and Middlefield. The seminar room is in Building A. Visitors should sign in at the reception desk in the Building A lobby. IMPORTANT: Visitors from Communist Bloc countries should make the necessary arrangements with Fran Leonard, SRI Security Office, (415) 859-4124, as soon as possible. ------------------------------ Date: Fri, 15 Jan 08:45:27 1988 From: dlm%research.att.com@RELAY.CS.NET Subject: Seminar - Four-Valued Semantics for Terminological Logics (AT&T) Title: A Four-Valued Semantics for Terminological Logics Speaker: Peter F. Patel-Schneider Schlumberger Palo Alto Research 3340 Hillview Ave. Palo Alto, California 94304 Date: Monday, January 18, 1988 Time: 10:30 AM Place: AT&T Bell Laboratories - Murray Hill 3D-473 Terminological logics formalize and extend the notions of concepts, roles, and restrictions present in semantic networks, frame-based systems, and object-oriented programming systems. The most important semantic relationship in these logics is subsumption-whether one concept is more general than another. Subsumption is a non-trivial relationship and if the terminological logic is expressively powerful, then determining whether one concept subsumes another is computationally intractable. Because of this intractability, knowledge representation systems based on terminological logics are not suitable for use in knowledge-based systems. This problem can be solved by using a four-valued semantics, resulting in an expressively powerful terminological logic which has tractable subsumption. The subsumptions supported by the logic are a type of "structural" subsumption, where each structural component of one concept must have an analogue in the other concept. Structural subsumption captures an important set of subsumptions, similar to the subsumptions computed in KL-ONE and NIKL. The four-valued semantics can thus be used to develop object-based knowledge representation systems suitable for use in knowledge-based systems. Sponsor: Ron Brachman ------------------------------ End of AIList Digest ********************