Date: Tue 7 Jun 1988 22:14-EDT From: AIList Moderator Nick Papadakis Reply-To: AIList@AI.AI.MIT.EDU Us-Mail: MIT Mail Stop 38-390, Cambridge MA 02139 Phone: (617) 253-2737 Subject: AIList Digest V7 #22 To: AIList@AI.AI.MIT.EDU Status: RO AIList Digest Wednesday, 8 Jun 1988 Volume 7 : Issue 22 Today's Topics: Queries: Response to: inductive expert system tools Stock Price Forecasting Response to: AI in weather forecasting Talk Announcement - "Mundane Reasoning" ---------------------------------------------------------------------- Date: 6 Jun 88 18:42:42 GMT From: esosun!cogen!alen@seismo.css.gov (Alen Shapiro) Subject: Response to: inductive expert system tools In article <402@dnlunx.UUCP> marlies@dnlunx.UUCP (Steenbergen M.E.van) writes: > > . I am engaged in artificial intelligence research. At the >moment I am investigating the possibilities of inductive expert systems. In >the literature I have encountered the names of a number of (supposedly) >inductive expert system building tools: Logian, RuleMaster, KDS, TIMM, >Expert-Ease, Expert-Edge, VP-Expert. I would like to have more information >about these tools (articles about them or the names of dealers in Holland). I >would be very grateful to everyone sending me any information about these or >other inductive tools. Remarks of people who have worked with inductive expert >systems are also very welcome. Thanks! > There are basically 2 types of inductive systems a) those that build an internal model by example (and classify future examples against that model) and b) those that generate some kind of rule which, when run, will classify future examples a) includes perceptron-like systems and more recently neural-net technology as well as some of the work my company does that is NOT neural-net based) b) may be split into 2 camps; 1) systems that produce a single decision tree for all decision classes (e.g. Quinlan's ID3 upon which RuleMaster, Expert-Ease, Ex-Tran, Superexpert, First Class and more are based); 2) systems that produce a decision for each class-value (e.g. Michalski's AQ11). I do not include those systems that are not able to generalise in either a or b since strictly they are not inductive!! I don't know about dealers in Holland but ITL at George House, 36 N. Hanover St., Glasgow Scotland G1 2AD (U.K.) are experts in producing REAL expert systems that are inductively derived. The Turing Institute (same address) are also well known in this regard. --alen the Lisa slayer (it's a long story) DISCLAIMER: I work for a company delivering inductively derived expert systems into the real world doing real work and saving real money. I can be counted on to be very biased!! ....!{seismo,esosun,suntan}!cogen!alen ------------------------------ Date: Tue, 7 Jun 88 15:52:30+0900 From: Minsu Shin Subject: Stock Price Forecasting I am looking for references (books, articles,...) or any information concerning "Forecast of Stock Price using Pattern-Recognition". I will produce the gathered information after receiving some amount of information, if anyone wants. Replies via email are fine. Many thanks in advance for this favor. My addresse is as follows: Network Intellegence Section ISDN Development Dept. ETRI P.O.Box 8, Tae-Deog Science Town Dae-Jeon,Chung-Nam, 302-350, KOREA Fax : 82-042-861-1033, Telex : TDTDROK K45532 ------------------------------ Date: Tue, 7 Jun 88 07:55:06 EDT From: m06242%mwvm@mitre.arpa Subject: Response to: AI in weather forecasting To: AILIST@AI.AI.MIT.EDU From: George Swetnam Subject: AI in Weather Forecasting In 1985, The MITRE Corporation and the National Center for Atmospheric Research collaborated in an experimental expert system for predicting upslope snowstorms in the Denver, Colorado area. An upslope storm is one which gets the necessary atmospheric lifting from translation of a moist airmass up a topographic slope. Upslope storms are responsible for roughly 60% of the precipitation in the Denver region; in this case the topographic slope is the slow, long rise from the Mississippi River to the foot of the Rocky Mountains. The most recent published information on this work is the paper whose title and abstract appear below. FIELD TRIAL OF A FORECASTER'S ASSISTANT FOR THE PREDICTION OF UPSLOPE SNOWSTORMS G. F. Swetnam and E. J. Dombroski, The MITRE Corporation R. F. Bunting, University Corporation for Atmospheric Research AIAA 25th Aerospace Sciences Meeting, January 12-15, 1987 Paper No. AIAA 87-0029 ABSTRACT An experimental expert system has been developed to assist a meteorologist in forecasting upslope snowstorms in the Denver, Colorado area. The system requests about 35 data entries in a typical session and evaluates the potential for adequate moisture, lifting, and cold temperatures. From these it forecasts the expected snowfall amount. The user can trace the reasoning behind the forecast and alter selected input data to determine how alternative conditions affect the expectation of snow. Written in Prolog, the system runs on an IBM PC or PC compatible microcomputer. A field trial was held in the winter of 1985-86 to test system operation and improve the rule base. The system performed well, but needs further refinement and automatic data collection before it can be considered ready for evaluation in an operational context. George Swetnam (gswetnam@mitre) The MITRE Corporation 7525 Colshire Drive McLean, VA 22102 Tel: (703) 883-5845 * * George :: ------------------------------ Date: Tue, 7 Jun 88 00:13:03 EDT From: research!dlm@research.att.com Subject: Talk Announcement ______________________________________________________________________ TALK ANNOUNCEMENT Speaker: Mark Derthick - Dept. of CS, Carnegie Mellon University Title: Mundane Reasoning Date: Tuesday, June 7 Time: 10:00 Place: AT&T Bell Laboratories MH 3D436 Abstract: Frames are a natural and powerful conception for organizing knowledge. Yet in most well-defined frame-based knowledge representation systems, such as KL-ONE, the knowledge base must be logically consistent, no guesses are made to remedy incomplete knowledge bases, and they sometimes fail to return answers in a reasonable time, even for seemingly easy queries. On the other hand are connectionist knowledge representation systems, which are more robust in that they can be made to always return an answer quickly, and knowledge is combined evidentially. Unfortunately these systems, if they have a well defined formal semantics at all, have had much less expressive power than symbolic systems. The differing characteristics result from two independent decisions. First, the statistical technique of Maximum a Posteriori estimation is used as a semantic foundation rather than logical deduction. Second, heuristic simplifications of the models considered give rise to fast, but errorful behavior. Having made this distinction, it is possible to use the same powerful syntax of symbolic systems, but interpret it statistically and implement it with a connectionist network. Although correct networks are exponentially large, they serve as a basis from which architectural simplifications can be made which preserve an intuitive connection to the formal theory. The knowledge base must be tuned to alleviate errors caused by the heuristic simplifications, so the system is intended for familiar everyday situations in which past performance has been used for training and in which the ramifications of wrong answers are not serious enough to justify the exponential search time required for provably correct behavior. Sponsor: Ron Brachman & Deborah McGuinness (allegra!dlm) ------------------------------ End of AIList Digest ********************