Date: Wed 23 Nov 1988 22:05-EST From: AIList Moderator Nick Papadakis Reply-To: AIList@AI.AI.MIT.EDU Us-Mail: MIT LCS, 545 Tech Square, Rm# NE43-504, Cambridge MA 02139 Phone: (617) 253-6524 Subject: AIList Digest V8 #132 To: AIList@AI.AI.MIT.EDU Status: R AIList Digest Thursday, 24 Nov 1988 Volume 8 : Issue 132 Queries: BOOTSTRAP translating LISP to/in other languages Prolog on a Macintosh II Input refutations OPS and Prolog comparison Responses: ES for Student Advising Genetic Learning Algorithms Learning arbitrary transfer functions (2 responses) Iterative Deepening (2 responses) AI & the DSM-III ---------------------------------------------------------------------- Date: Wed, 16 Nov 88 20:21:35 EST From: "Thomas W. Stuart" Subject: BOOTSTRAP I'm passing along a query from Dr. William McGrath, here at the School of Information and Library Studies, SUNY - Buffalo. He is looking for references or information about available programs and packages for Efron's Bootstrap statistical procedures -- packages which might run on micros or VAX systems. ------------------------------ Date: 21 Nov 1988 07:55:54 CDT From: Walter.Daugherity@LSR.TAMU.EDU Subject: translating LISP to/in other languages I am looking for information about converting LISP to other languages (C, PASCAL, ADA, etc.) or about LISP interpreters written in such languages. Thanks in advance, Walter Daugherity ARPA INTERNET: daugher@cssun.tamu.edu Walter.Daugherity@lsr.tamu.edu CSNET: WCD7007%LSR%TAMU@RELAY.CS.NET WCD7007%SIGMA%TAMU@RELAY.CS.NET BITNET: WCD7007@TAMLSR WCD7007@TAMSIGMA ------------------------------ Date: Tue, 22 Nov 88 14:33:50 +1000 From: "ERIC Y.H. TSUI" Subject: Prolog on a Macintosh II I would like to communicate with users of the following PROLOGs on a MAC: M-1.15 (from Advanced AI Systems Prolog) IF/Prolog (from Interface Computer Gmbh.) Prolog-1 (from Expert Systems International) ZYX Macintosh Prolog 1.5 (from ZYX Sweden AB) (Welcome any other suggestions for Prolog on a MAC II ?) I am porting a large (approx. 1MB source) MU-Prolog (almost exactly DEC-10 Edinburgh Prolog syntax) system to run on a MAC II. Desirable features include: save states, no need to pre-declare dynamic predicates (flexible assert and retract), reconsult, large stack space and efficient execution. Eric Tsui eric@aragorn.oz Research Associate Department of Computing and Mathematics Deakin University Geelong, Victoria 3217 AUSTRALIA ------------------------------ Date: 22 Nov 88 07:46:51 GMT From: geoff@wacsvax.OZ (Geoff Sutcliffe) Subject: Input refutations I have been searching (in the wrong places obviously) for a proof that resolution & paramodulation, or resolution & paramodulation & factoring, form a complete input refutation system for sets of Horn clauses, and that the single negative clause in a minimally unsatisfiable set of Horn clauses may be used as the top clause in such refutations. Refutation completeness, without specification of the top clause, is in "Unit Refutations and Horn Sets" [Henschen 1974]. If set-of-support is compatible with input resolution,paramodulation,factoring then it is possible to choose the negative clause as the support set, and the problem is solved. Is this compatibility known? Any help, with this seemingly obvious result, would be appreciated. Geoff Sutcliffe Department of Computer Science, CSNet: geoff@wacsvax.oz University of Western Australia, ARPA: geoff%wacsvax.oz@uunet.uu.net Mounts Bay Road, UUCP: ..!uunet!munnari!wacsvax!geoff Crawley, Western Australia, 6009. PHONE: (09) 380 2305 OVERSEAS: +61 9 380 2305 ------------------------------ Date: 22 Nov 88 16:02:20 GMT From: att!whuts!homxb!hou2d!shun@bloom-beacon.mit.edu (S.CHEUNG) Subject: OPS and Prolog comparison I am looking for some information comparing OPS83 (including OPS5 and C5) and Prolog, such as speed, the types of applications they are good for, availability, ease of software maintenance, how easy to learn, etc. I am also interested in statistics concerning the number of existing applications using each language. There might be articles on these topics already; can someone let me know where to find them? Thanks in advance. -- Shun Cheung -- -- Shun Cheung, AT&T Bell Laboratories, Middletown, New Jersey electronic: shun@hou2d.att.com or ... rutgers!mtune!hou2d!shun voice: (201) 615-5135 ------------------------------ Date: Thu, 17 Nov 1988 20:22:35 EST From: "Thomas W. Stuart" Subject: ES for Student Advising William McGrath (School of Information and Library Studies, 309 Baldy, SUNY at Buffalo, 14120) has created an ES knowledgebase (KB) for advising students on what courses to take for a projected plan of study in library and information science, particularly in reference to the student's specific career objective. The KB, created with 1stCLASS ES shell, considers the type of job environment (academic, public, corporate, sci-tech) and type of work (collection development, cataloging, information retrieval, management, etc.), prerequisites, hours needed to complete the program, need for faculty permission, and other factors. Planned modules: advice for resolving schedule conflicts, list of job prospects -- given the student's program, feedback and evaluation. ------------------------------ Date: 7 Nov 88 12:03:04 GMT From: mcvax!ukc!strath-cs!pat@uunet.uu.net (Pat Prosser) Subject: Re: GENETIC LEARNING ALGORITHMS Genetic Algorithms (GA's) traditionally represent the genetic string (chromosone) using a binary alphabet; Holland has shown this to be optimal. It is not the only alphabet, a purely symbolic alphabet is possible if appropriate genetic operators are defined. For example [1] P. Prosser, "A Hybrid Genetic Algorithm for Pallet Loading" European Conference on Artificial Intelligence, 1988 [2] Derek Smith, "Bin Packing with Adaptive Search" Proceedings ICGAA 1985 [3] David Goldberg, "Alleles, Loci and the Travelling Salesman Problem" The only problem with non-binary alphabet is the limits of our imagination. ------------------------------ Date: Fri, 18 Nov 88 10:18:37 EST From: alexis%yummy@gateway.mitre.org Reply-to: alexis%yummy@gateway.mitre.org Subject: Flaming on Neural Nets and Transfer Functions I have to admit some surprise that so many people got this "wrong." Our experience is that neural nets of the PDP/backprop variety are at their *BEST* with continueous mappings. If you just want classification you might as well go with nearest-neighbor alg.s (or if you want the same thing in a net try Nestor's Coulombic stuff). If you can't learn x=>sin(x) in a couple of minutes, you've done something wrong and should check your code (I'm assuming you thought to scale sin(x) to [0,1]). Actually, requiring a PDP net to output 1's and 0's means your weights must be quite large which takes alot of time and puts you way out on the tails of the sigmoids where learning is slow and painful. What I do for fun (?) these days is try to make nets output sin(t) {where t is time} and other waveforms with static or "seed" wave inputs. For those who like math, G. Cybenko (currently of U. Illinois and starting 12/10/88 of Tufts) has a very good paper "Approximation by Superpositions of a Sigmoidal Function" where he gives a existence proof that you can uniformly approximate any continuous function with support in the unit hypercube. This means a NN with one hidden layer (1 up from a perceptron). Certainly more layers generally give more compact and robust codings ... but the theory is *finally* coming together. Alexis Wieland .... alexis%yummy@gateway.mitre.org ------------------------------ Date: 17 Nov 88 20:48:52 GMT From: amos!joe@sdcsvax.ucsd.edu (Shadow) Subject: Re: Learning arbitrary transfer functions in article, 399.uvaee.ee.virginia.EDU writes: >>I am looking for any references that might deal with the following >>problem: >> >>y = f(x); f(x) is nonlinear in x >> >>Training Data = {(x1, y1), (x2, y2), ...... , (xn, yn)} >> >>Can the network now produce ym given xm, even if it has never seen the >>pair before? >> >>That is, given a set of input/output pairs for a nonlinear function, can a >>multi-layer neural network be trained to induce the transfer function my response: 1. Neural nets are an attempt to model brain-like learning (at least in theory). So, how do human's learn non linear functions ? : you learn that x^2, for instance, is X times X. And how about X times Y ? How do humans learn that ? : you memorize it, for single digits, and : for more than a single digit, you multiply streams of digits together in a carry routine. 2. So the problem is a little more complicated. You might imagine a network which can perfectly learn non-linear functions if it has at its disposal various useful sub-networks (e.g., a network can learn x^n if it has at its disposal some mechanism and architecture suitable for multiplying x & x.) (imagine a sub-network behaving as a single unit, receiving input and producing output in a predictable mathimatical manner) (promoting thought) What is food without the hunger ? What is light without the darkness ? And what is pleasure without pain ? joe@amos.ling.ucsd.edu ------------------------------ Date: Fri, 18 Nov 88 19:50:27 pst From: purcell%loki.edsg@hac2arpa.hac.com (ed purcell) Subject: iterative deepening for game trees, state-space graphs Some observations on the request of quintus!ok@unix.sri.com (16 Nov 88) for references on the term ``iterative deepening'': In his IJCAI85 paper on the IDA* (Iterative Deepening A*) search algorithm for state-space problem graphs, Rich Korf of UCLA acknowledges early chess-playing programs as the first implementations of the idea of progressively deeper searches. (The history of progressively deeper look-ahead searches for game trees is somewhat reminiscent of the history of alpha-beta pruning -- these clever algorithms were both implemented early but not immediately published nor analyzed until many years later.) The closely-related term ``progressive deepening'' also has been around awhile; for example, this term is used in the 2nd edition (1984) of Pat Winston's textbook ``An Introduction to AI.'' The contributions of Korf's IJCAI85 paper on IDA* are in the re-formulation and analysis of progressively deeper depth-first search for state-space graphs, using a heuristic evaluation function instead of a fixed depth bound to limit node expansions. It is interesting that Korf is now investigating the re-formulation of minimax/alpha-beta pruning for state-space graphs. Ed Purcell purcell%loki.edsg@hac2arpa.hac.com 213-607-0793 ------------------------------ Date: 19 Nov 88 02:16:34 GMT From: korf@locus.ucla.edu Subject: "Iterative-Deepening" Reference wanted Another reference on this subject is: "An analysis of consecutively bounded depth-first search with applications in automated deduction", by Mark E. Stickel and W. Mabry Tyson, in IJCAI-85, pp. 1073-1075. ------------------------------ Date: 23 Nov 88 18:07:05 GMT From: sire@ptsfa.PacBell.COM (Sheldon Rothenberg) Subject: Re: AI & the DSM-III In a previous article, ANDERSJ%ccm.UManitoba.CA@MITVMA.MIT.EDU writes: > Hi Again. I have a colleague who is attempting to write a paper on > the use of AI techniques in psychiatric diagnosis in general, and > more specifically using the DSM-III. Todd Ogasawara, at U. of Hawaii, posted a 10 article biblio on related topics. The article which appears most relevant is: Hardt, SL & MacFadden, DH Computer Assisted Psychiatric Diagnosis: Experiments in Software Design from "Computers in Biology and Medicine", 17, 229-237 A book by DJ Hand entitled "Artificial Intelligence and Psychiatry" a 1985 publication of Cambridge University Press also looks promising. Todd's e-mail address on INTERNET is: todd@uhccux.UHCC.HAWAII.EDU Shelley Rothenberg (415) 867-5708 ------------------------------ End of AIList Digest ********************