[Prev][Next][Index][Thread]

CSLI Calendar, 18 March 1998, vol. 13:25



   
     C S L I   C A L E N D A R   O F   P U B L I C   E V E N T S
______________________________________________________________________

18 March 1998                   Stanford               Vol. 13, No. 25
______________________________________________________________________

                     A weekly publication of the
       Center for the Study of Language and Information (CSLI)
      Stanford University, Ventura Hall, Stanford, CA 94305-4115
                             ____________

             ACTIVITIES DURING 18 MARCH TO 27 MARCH 1998

WEDNESDAY, 18 MARCH
         3:15pm ME297: Design Theory and Methodology Seminar
                Grammatical Generation and Optimizing Search in Early
                Design
                Professor Jonathan Cagan
                George Tallman and Florence Barrett Ladd Development
                Professor in Engineering
                Department of Mechanical Engineering,
                Carnegie Mellon University  
                (this may be a closed seminar, check the web page)
                http://cdr.stanford.edu/DD/Courses/me297/
                Abstract below
     
THURSDAY, 19 MARCH
        10:00am Neurobiology Seminar
                Beyond the M and P Pathways: How Early Neuronal Diversity
                Contributes to Visual Perception
                Dr. David Calkins
                The Zanvyl Krieger Mind/Brain Institute
                The Johns Hopkins University
                Host: Dr. Eric Knudsen
                http://www.stanford.edu/dept/nbio/seminars.html  

        12 noon CSLI Talk
                Cordura Hall, Room 100
                Global Intelligence Project
                Prof. Masa Numao
                Tokyo Institute of Technology
                Abstract below
            
         2:15pm Semantics Workshop
                Margaret Jacks Hall 460:146
                Frank Veltman
                University of Amsterdam and the Department of Philosophy,
                Stanford University
                Abstract below
                
         4:15pm Seminar on Computational Learning and Adaptation (SCLA)
                Gates 100
                Learning Evaluation Functions for Optimization
                Justin Boyan 
                Carnegie Mellon University, Pittsburgh, PA
                Abstract below

FRIDAY, 20 MARCH
        12 noon Logic Lunch
                Room 380:383N
                Reductions of Theories of Countable Tree Ordinals to ID_1
                Sol Feferman
                Abstract below

THURSDAY, 26 MARCH
        12 noon CSLI Talk
                Cordura 100
                Grounding Meaning through Language Games
                Prof. Luc Steels
                VUB AI Lab (Brussels)
                and Sony CSL (Paris)
                Abstract below

         4:15pm Seminar on Computational Learning and Adaptation (SCLA)
                Gates 100
                Summarizing Similarities and Differences Among Related
                Documents
                Eric Bloedorn
                Mitre, McLean, VA.
                Abstract below

FRIDAY, 27 MARCH
        10:00am Neurobiology Seminar
                Neural Learning Rules in the Cerebellum: Insights from
                the Vestibulo-ocular Reflex
                Dr. Jennifer Raymond
                Department of Physiology and Keck Center for Integrative
                Neuroscience
                University of California, San Francisco
                Host: Dr. Eric Knudsen
                http://www.stanford.edu/dept/nbio/seminars.html  
                             ____________

                            SPRING QUARTER

                            Philosophy 169
                          Intensional Logic
                          Johan van Benthem
                   Tuesday, Thursday, 1:15pm-2:30pm
                             Bldg. 60:62A
                        Starting 31 March 1998

We do a basic introduction to modal logic, in a modern perspective
using bisimulation, first-order correspondence, and other recent tools
that are not yet standard textbook fare. After that, we'll look at
some active areas of applications, including dynamic and epistemic
logics, plus some modal logics for grammar formalisms.  Finally, some
guest speakers will highlight recent perspectives (metaphysics of
modality, computational complexity, programming).  At the end of the
course, students should be able to understand research papers that use
modal techniques, and have an intelligent appraisal of what makes them
tick.

                            Philosophy 298
              Topics in Logic, Language, and Computation
                          Johan van Benthem
                        Friday, 1:15pm-3:05pm
                             Bldg. 60:61F
                        Starting 3 April 1998

This year's theme is *Games in Logic*. Games are an upcoming paradigm
for dealing with dynamic aspects of action and information flow, which
actually revives a very old tradition in the field. In addition to the
received syntactic and semantic views on validity, there is a
'pragmatic' intuition which says that the valid inferences are
precisely those that correspond to winning strategies in an
argumentation game. Logical games have never entered the core of the
field, partly because no Hilbert, Tarski, or Turing has yet provided
them with a definitive conceptual analysis from first principles. We
will survey some important games (due to Ehrenfeucht-Fraisse,
Lorenzen, Hintikka, and others) and their modern uses (including
connections with linear logic). Then we look at some general
mathematical theory dealing with game rules and strategies.  Finally,
we look into recent connections between game theory as practiced by
economists and philosophers studying rationality, and
epistemic-dynamic logics that place action and information flow at
center stage (cf. J. van Benthem, "Exploring Logical Dynamics", CSLI
Publications, Stanford, 1996).
                             ____________

             ME297: DESIGN THEORY AND METHODOLOGY SEMINAR
                 on Wednesday, 18 March 1998, 3:15pm
              http://cdr.stanford.edu/DD/Courses/me297/
               (seminar may be closed, check web page)

     Grammatical Generation and Optimizing Search in Early Design
                            Jonathan Cagan
               George Tallman and Florence Barrett Ladd
                 Development Professor in Engineering
                 Department of Mechanical Engineering
                      Carnegie Mellon University
                                  
The early stages of design require search through large spaces of
design concepts. To keep the process tractable and productive, the
description of the design space must be concise, yet the scope of the
space still vast. Further, once a design space is defined it must be
searched in an efficient manner to determine good design solutions. In
this presentation we will explore the benefits of shape grammars to
describe engineering design spaces. These grammars either model
knowledge intensive products such as coffee makers, or
near-knowledge-free configurations such as truss structures. To search
design spaces such as those generated by knowledge-free grammars,
techniques must be able to move through discontinuous, non-smooth
topologies and models and determine not only feasible solutions but
optimally directed ones. We will explore how stochastic optimization
methods find optimal configurations in these spaces. One such
application will be in the area of structural topology design and
another in the area of product layout. We will also briefly examine
the relationship between these search methods and cognitive models of
human problem solving.
                             ____________

                              CSLI TALK
                 on Thursday, 19 March 1998, 12 noon
                        Cordura Hall, Room 100
             http://www-csli.stanford.edu/csli/Coglunch/

                     Global Intelligence Project
                              Masa Numao
               Professor, Tokyo Institute of Technology
                  http://www.titech.ac.jp/home.html

We propose a cooperative system of human beings and computers, called
"Global Intelligence" (GI). Human workers in GI are supported by a
computer system called GIANT (GI Associating NeTwork), which consists
of an inference mechanism by using a dynamically transforming network,
and has a learning mechanism by adjusting a weight on each link in the
network. After showing their structures and experiments, we point out
that we can construct an interesting model of intelligence not only by
analyzing a "Society of Mind", but also by synthesizing "Mind of a
Society".

A limited number of sandwiches available for purchase at the talk.
                             ____________

                     STANFORD SEMANTICS WORKSHOP
                  on Thursday, 19 March 1998, 2:15pm
                     Margaret Jacks Hall 460:146
      http://www-linguistics.stanford.edu/Linguistics/semgroup/

              Subjunctive Modalities and subordination.
                            Frank Veltman
                       University of Amsterdam
        and the Department of Philosophy, Stanford University

I will sketch some of the problems one encounters if one tries to
develop a dynamic semantics for subjunctive modalities like `might',
`would', `might have been' and `would have been', as these occur not
only in subjunctive conditionals but also in other kinds of discourse.
Taking subjunctive conditionals as my starting point, I will among
other things argue that the dynamic theories developed so far
(Barwise, Morreau) are wrong in that they predict that any agent who
at a given time is entitled to entertain the indicative conditionals
`If A is the case, then so is C' will later, after learning that A is
in fact false, be entitled to entertain the counterfactual `If A had
been the case, C would have been the case'. I will sketch an
alternative theory, show how it solves some notorious puzzles, and
explain how it can be extended to a theory that can explain, for
instance, why a text like: "I didn't drink any wine. It would have
made me sick" is easy to understand whereas "I drank some wine. It
would not have made me sick" is incoherent.

                             ____________
   
       SEMINAR ON COMPUTATIONAL LEARNING AND ADAPTATION (SCLA)
             on Thursday, 19 March 1998, 4:15pm to 5:30pm
                              Gates 100
              http://www-csli.stanford.edu/cll/scla.html

            Learning Evaluation Functions for Optimization
                             Justin Boyan
                    mailto:jab@lefty.sp.cs.cmu.edu
       
STAGE is a new search technique which learns a problem-specific
heuristic evaluation function as it searches. The heuristic is trained
by least-squares TD(lambda) to predict, from features of states along
the search trajectory, how well a fast Markovian search method such as
hill-climbing will perform starting from each state. Search proceeds
by alternating between two stages: performing the fast search to
gather new training data, and following the learned heuristic to reach
a promising new start state.

STAGE has produced good results on a variety of combinatorial
optimization domains, including VLSI channel routing, Bayes net
structure-finding, bin-packing, Boolean satisfiability, radiotherapy
treatment planning, and geographic cartogram design. I'll discuss as
many of these successes as time permits, and also explain a STAGE
failure on the domain of inverse Boggle.
                             ____________

                             LOGIC LUNCH
                  on Friday, 20 March 1998, 12 noon
                         Math Corner 380:383N
              http://www-philosophy.stanford.edu/Logic/

      Reductions of Theories of Countable Tree Ordinals to ID_1.
                             Sol Feferman

This is the last logic lunch of the quarter.  We plan an informal
lunch at the coffee house afterward.

This is a report of a section from the forthcoming chapter by Jeremy
Avigad and myself on Goedel's functional ("Dialectica"
interpretation.* I will sketch two applications of this
interpretation, one for a classical and one for an intuitionistic
theory of countable tree ordinals, to reduce them to corresponding
systems of one arithmetical inductive definition.  The questions
following this work are: (i) how these are related, and (ii) can
anything similar be done for theories of higher ordinal number
classes?

*To appear in the Handbook of Proof Theory, S. Buss ed.
                             ____________

                              CSLI TALK
                 on Thursday, 26 March 1998, 12 noon
                        Cordura Hall, Room 100
             http://www-csli.stanford.edu/csli/Coglunch/

               Grounding Meaning through Language Games
                           Prof. Luc Steels
                        VUB AI Lab (Brussels)
                      http://arti.vub.ac.be/www/
                         and Sony CSL (Paris)
                       http://www.csl.sony.fr/

Experiments are presented with physical robotic agents operating in a
real-world environment. It is shown how these agents can build up a
repertoire of perceptually grounded distinctions through
discrimination games and how they can develop a shared lexicon
verbalising these distinctions through adaptive naming games.
       
These experiments illustrate that language can be viewed as a complex
adaptive system that emerges through self-organization from the local
interactions of individuals. Language continues to evolve and adapt
due to stochasticity (which introduces innovation), uncertainty (which
maintains variation), and constant renewal of its user population.
       
Biography: Luc Steels is a professor in Artificial Intelligence at the
University of Brussels (VUB). He also directs the Sony Computer
Science Laboratory in Paris. His current work focuses on
behavior-oriented approaches to sensori-motor intelligence using
physical robotic agents as experimental platforms, and the origins and
evolution of language.

A limited number of sandwiches available for purchase at the talk.
                             ____________
   
       SEMINAR ON COMPUTATIONAL LEARNING AND ADAPTATION (SCLA)
             on Thursday, 26 March 1998, 4:15pm to 5:30pm
                              Gates 100
              http://www-csli.stanford.edu/cll/scla.html

   Summarizing Similarities and Differences Among Related Documents
                           Eric E. Bloedorn
                          MITRE Corporation
                   mailto:bloedorn@azrael.mitre.org
     
Text summarization attempts to address the information overload
problem by taking a partially-structured source text, extracting
information content from it, and presenting the most important content
to the user in a manner sensitive to the user's or application's
needs. The first part of the talk will describe WebSumm, a system for
summarizing related documents. The approach in WebSumm exploits recent
progress in information extraction to represent salient units of text
and their relationships. By exploiting meaningful relations between
units based on an analysis of text cohesion and the context in which
the comparison is desired, the summarizer can pinpoint similarities
and differences, and align text segments. The second part of the talk
will describe an application of machine learning methods to train our
summarizer. The goal of this learning approach is to have a system
capable of adjusting summarizers to better fit the user's interest.
                             ____________

                             END MATERIAL

The CSLI Calendar appears weekly on Wednesdays throughout the academic
year.  Announcements, abstracts, and other information to appear in
the Calendar should be submitted to the editor, who reserves the right
to decide what does or does not go in the calendar
mailto:incalendar@csli.stanford.edu.

Requests to be added to the mailing list should be sent to
mailto:majordomo@csli.stanford.edu.  With the lines in the body of
the text of either
        subscribe csli-calendar
for the long form or
        subscribe csli-short-calendar
for the short form.  Problems with subscribing or unsubscribing should
be sent to owner-csli-calendar@csli.stanford.edu. 

The full current issue is at
http://www-csli.stanford.edu/csli/Archive/calendar/current.html
and the archives at
http://www-csli.stanford.edu/csli/Archive/calendar/.  

People on most of the CSLI computers can type 'help csli-calendar' to
see the current issue.  They can also type 'help quip' to see the
Linguistics Department newsletter, the Sesquipedalian,
http://www-linguistics.stanford.edu/Linguistics/.

The CSLI Calendar is also posted each week to
news://nntp-csli.stanford.edu/csli.bboard.  

Information about CSLI's research program is available at
http://www-csli.stanford.edu/csli/.

For maps to the Stanford University campus see
http://www.stanford.edu/home/visitors/maps.html.
                             ____________