Date: Wed 15 Jun 1988 23:24-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 #35 To: AIList@AI.AI.MIT.EDU Status: RO AIList Digest Thursday, 16 Jun 1988 Volume 7 : Issue 35 Today's Topics: Queries: Connectionist Expert Systems Seminar: Supercomputing Summer Institute Philosophy: Biological relevance and AI Human-human communication ---------------------------------------------------------------------- Date: 14 Jun 88 18:22:56 GMT From: ndsuvax!ncthangi@uunet.uu.net (sam r. thangiah) Subject: Connectionist Expert Systems >From: olivier@boulder.Colorado.EDU (Olivier Brousse) >Could any one give me pointers about NESTOR, an expert system combining >neural nets and symbolic AI techniques ? >Is there any other work done in connectionist expert systems ? I am also interested in the above area. Pointers to any references, papers or work being conducted using such techniques will be much appreciated. Thanks in advance, Sam -- Sam R. Thangiah, North Dakota State University. UUCP: ...!uunet!ndsuvax!ncthangi BITNET: ncthangi@ndsuvax.bitnet ARPA,CSNET: ncthangi%ndsuvax.bitnet@cunyvm.cuny.edu ------------------------------ Date: Wed, 15 Jun 88 11:09:55 EDT From: hendler@dormouse.cs.umd.edu (Jim Hendler) Subject: Re: Connectionist expert systems There are essentially two strains of work going on in this area. One strain involves the actual implementation of expert-system-like programs using connectionist techniques. I'll let the real connectionists comment on those. The second strain involves the creation of hybrid systems which have both connectionist and symbolic mechanisms cooperating in the solution of traditional cognitivish AI problems (i.e. language, planning, ``high level'' vision, etc.) As well as my own work in this area, there is the work of Wendy Lehnert of UMass, Michael Dyer of UCLA, Mark Jones of Bell Labs, and, depending on how you classify it, the work of Dave Touretzky at CMU (Dave's work doesn't really have symbols in the traditional sense of the word, but some of his models use gating and other serial techniques to control a larger connectionist system). Also, the work of the local (now often called structured) connectionists have in some ways been hybrid. The one most related might be that of Lokendra Shastri of U Penn. -Jim Hendler U. of Md. Institute for Advanced Computer Studies UMCP College Park, Md. 20742 ------------------------------ Date: Wed, 15 Jun 88 10:18:09 EDT From: Una Smith Subject: Supercomputing Summer Institute John von Neumann National Supercomputer Center 1988 Summer Institute "An Intensive Introduction to Vector and Parallel Supercomputing" August 1-12, 1988 Objectives The John von Neumann National Supercomputer Center (JvNC) will hold its third annual Summer Institute during August 1-12, 1988 at the JvNC in Princeton, New Jersey. The principal goals of the Institute are to teach the participants how to use supercomputers effectively and to provide an intensive mini-course in computational scientific research. Facilities The JvNC operates the first production Class VII supercomputer, the ETA 10, installed at the JvNC in March, 1988. The ETA 10 is presently configured with 128 million words of shared memory and four Central Processing Units (CPUs) each equipped with four million words of local memory. The JvNC also operates two CYBER 205 supercomputers, configured with four million words of memory. All supercomputers are accesssed from the DEC 8600 front-end computers at the JvNC operating under VMS or Ultrix (UNIX). The PEP software, a series of user-friendly, command driven interactive procedures, allows users to execute programs on the ETA 10 and CYBER 205 without the need for learning the supercomputer's operating system. The JvNC Visualization Facility includes two Silicon Graphics Iris 4D workstations and two Sun 3/160 workstations with presentation and draft quality color hardcopy. Operation The Summer Institute will be equally divided between the classroom and the laboratory. The classroom portion will include lectures by JvNC staff on the overall JvNC computer environment, vector computing, parallel computing, advanced performance programming, communications and networking, and graphics. In addition, there will be invited lectures on computational chemistry, algorithms for parallel machines, computational fluid dynamics, plasma physics, and visualization. The laboratory portion will include detailed consulting by User Services on program optimization for the ETA 10 and CYBER 205, and utilization of the JvNC Visualization Facility. The participants shall be awarded an allocation of supercomputer time for their research projects. Participants Postdoctoral, graduate and advanced undergraduate students at U.S. universities and colleges are eligible to apply for the Summer Institute. Knowledge of Fortran is a prerequisite. Experience with vectorization and supercomputers is not required. Each participant is expected to be engaged in a computational research project and bring an operational Fortran code for further development and production runs during the Institute. Financial Support The JvNC will provide reimbursement for travel, accommodations and meals up to $1500 per participant according to NSF policies. Air travel shall be economy class and local accommodations will be organized through the JvNC. Meals will be reimbursed up to a maximum per diem. There is no separate stipend. Application Interested persons should submit the following information: - Curriculum vitae, indicating university or college affiliation, courses of study, undergraduate and graduate transcript (as appropriate), address and telephone (office and home) - Two letters of recommendation from faculty advisors or instructors - Description of computational research project The information should be submitted by 24 June 1988. All applications should be sent to: John von Neumann National Supercomputer Center Summer Institute P.O. Box 3717 Princeton, NJ 08543 For further information, contact David Salzman at 609/520-2000 or or ------------------------------ Date: 14 Jun 88 20:57:06 GMT From: tektronix!sequent!mntgfx!msellers@bloom-beacon.mit.edu (Mike Sellers) Subject: Biological relevance and AI (was Re: Who else isn't a science?) In article <13100@shemp.CS.UCLA.EDU>, Benjamin Thompson writes: >In article <10510@agate.BERKELEY.EDU> weemba@garnet.berkeley.edu writes: >> Gerald Edelman, for example, has compared AI with Aristotelian >> dentistry: lots of theorizing, but no attempt to actually compare >> models with the real world. AI grabs onto the neural net paradigm, >> say, and then never bothers to check if what is done with neural >> nets has anything to do with actual brains. Where are you getting your information regarding AI & the neural net paradigm? I agree that there is a lot of hype right now about connectionist/neural nets, but this alone does not invalidate them (they may not be a panacea, but they probably aren't worthless either). There are an increasing number of people interested in (and to some degree knowledgeable of) both the artificial and biological sides of sensation, perception, cognition, and (some day) intelligence. See for example the PDP books or Carver Mead's upcoming book on analog VLSI and neural systems (I just finished a class in this -- whew!). There have been recent murmurings from some of the more classical AI types (e.g. Seymour Papert in last winter's Daedalus) that the biological paradigm/metaphor is not viable for AI research, but these seem to me to be either overstating the case against connectionism or simply not aware of what is being done. Others contend that anything involving 'wetware' is not *really* AI at all, and thus shouldn't invade discussions on that subject. This is, I believe, a remarkably short-sighted view that amounts to denying the possibility of a new tool to use. > This is symptomatic of a common fallacy. Why should the way our brains > work be the only way "brains" can work? Why shouldn't *A*I workers look > at weird and wonderful models? We (basically) don't know anything about > how the brain really works anyway, so who can really tell if what they're > doing corresponds to (some part of) the brain? > > Ben I think Ben's second and following sentences here are symptomatic of a common fallacy, or more precisely of common misinformation and ignorance. No one has said or implied that biological nervous systems have a monopoly on viable methodologies for sensation, perception, and/or cognition. There probably are many different ways in which these types of problems can be tackled. We do have a considerable amount of knowledge about the human brain, and (for the time being more to the point) about invertebrate nervous systems and the actions of individual neurons. And finally, correspondence to biological systems, while important, is by no means a single and easily acheived goal (see below). On the other hand, we can say at least two things about the current state of implemented cognition: 1) The methods we now call 'classical' AI, starting from about the late 1950's or early 60's, have not made an appreciable dent in their original plans nor even lived up to their original claims. To refresh your memory, a quote from 1958: "...there are now in the world machines that think, that learn and that create. Moreover, their ability to do these things is going to increase rapidly until --in a visible future-- the range of problems they can handle will be coextensive with the range to which the human mind has been applied." This quote is from H. Simon and A. Newell in "Heuristic Problem Solving: The Next Advance in Operations Research" in _Operations Research_ vol 6, published in *1958*. (It was recently quoted by Dreyfus and Dreyfus in the Winter 1988 edition of Daedalus, on page 19.) We seem to be no closer to the realization of this claim than we were thirty years ago. 2) We do have one instance that proves that sensation, perception, and cognition are possible: natural nervous systems. Thus, even though there may be other ways of solving the problems associated with vision, for example, it would seem that adopting some of the same strategies used by other successful systems would increase the likelyhood of our success. While it is true that there is more unknown than known about nervous systems, we do know enough about neurons, synapses, and small aggregates of neurons to begin to simulate their structure and function. The issue of how much to simulate is a valid and interesting one. Natural nervous systems have had many millions of years to evolve their solutions (much longer than we hope to have to take with our artificial systems), but then they have been both undirected in their evolution and constrained by the resources and techniques available to biological systems. This would seem to argue for only limited biological relevance to artificial solutions: e.g., where neurons have axons, we can simply use wires. On the other hand, natural systems also have the tendency to take a liability and make it into a virtue. For example, while axons are not simple 'wires', and in fact are much slower, larger, and more complex than wires, they can also act as active parts of the whole system, enabling such things as temporal differentiation to occur easily and without much in the way of cellular overhead. Thus, while we probably will not want to create fully detailed simulations of neurons, synapses, and neural structures, we do need to understand what advantages are embodied in the natural approach and extract them for use in our artifices while carefully excluding those things that exist only by being carried along with the main force of the evolutionary current. All of this is not to say that AI researchers shouldn't look at "weird and wonderful models" of perception and cognition; this is after all precisely what they have been doing for the past thirty years. The only assertion here is that this approach has not yielded much in the way of fertile results (beyond the notable products such as rule-based systems, windowed displays, and the mouse :-) ), and that with new technology, new knowledge of biological systems, and a new generation of researchers, the one proven method for acheiving real-time sensation, perception, and cognition ought to be given its chance to fail. Responses welcomed. -- Mike Sellers ...!tektronix!sequent!mntgfx!msellers Mentor Graphics Corp., EPAD msellers@mntgfx.MENTOR.COM "AI is simply the process of taking that which is meaningful, and making it meaningless." -- Tom Dietterich (admittedly, taken out of context) ------------------------------ Date: Wed, 15 Jun 88 09:20:07 pdt From: Ray Allis Subject: Re: Human-human communication In AIList Digest V7 #31, Stephen Smoliar writes: > First of all, NO dance notation provides sufficient information for the > exact reproduction of a movement. Likewise there's not sufficient information in an English description of "red" to impart knowledge to a listener. > Ultimately, I tend to agree with Gilbert that the problem is not in the > notation but in what is trying to be communicated. Video is as valuable > in reconstructing dances as it is in gymnastics, but there is still no > substitute for "shaping" bodies. What Gilbert calls "memory positions" > I have always called "muscular memory;" and I'm afraid there is no substitute > for physical experience when it comes to acquiring it. Your experience with dance notation is illustrative of a characteristic of languages in general, and a seriously flawed assumption in "AI". "Natural" language *evokes* experience in a listener; language can't *impart* experience. No amount of English description will produce the experience of "red" in a congenitally blind person, or a computer, or produce the same quality of associations with "flame" and "danger" and "hot" and "blushing" that a sighted person can hardly avoid. In order for a computer (read digital computer) to "understand" human language, it must have *experience* which the language can evoke. "Data structures" won't do, because they are symbols themselves, not experience. In iconic languages, (e.g. the dance notations you mention) there is a small amount of information conveyed because the perception of the icon itself is an experience. Seeing a picture of a platypus is similar to seeing a platypus. Reading or listening to a description of a platypus is not. Hearing "cerulean" described in English conveys no information, and any "understanding" on the part of the receiver must be *created* from that receiver's experience. It is this line of thought which led me several years ago to discard the Physical Symbol System Hypothesis. Physical symbol systems are *not* sufficient to explain or reproduce human thought and behavior. They are formal systems (form-al: concerning form, eliminating content). The PSSH is, however, a useful guide for most of what is called "AI", which is the mechanization of formal logic, an engineering task and properly a part of computer science. That task has nothing to do with the creation of intelligence. I can certainly understand the irritation of the engineers at people who want to re-think such a job after it's started. ------------------------------ End of AIList Digest ********************