|
|
In addition to the three invited talks, we are delighted to announce
the following tutorial:
From Reality to Databases: a One-to-Many Relationship
Stefano
Spaccapietra
Swiss Federal Institute of Technology in Lausanne
(EPFL)
CH-1015 Lausanne, Swizerland
Stefano.Spaccapietra@epfl.ch
This tutorial focuses on data modeling abstractions that are needed
for an accurate conceptual design of traditional (alphanumeric)
databases as well as spatio-temporal databases. We first recall the
basic steps and concepts in a conceptual design approach. This leads
to schemas that describe the real world data in terms of complex
objects and relationships, properties and is-a links. Next we further
investigate issues related to supporting multiple representations for
the same real world entity. We show limitations in this respect of
current object-oriented data modeling approaches and propose solutions
to overcome such limitations.
Distinguised Invited Speakers
July 27, 2000: Professor Thomas
Dietterich (AAAI Fellow), Oregon State University.
July 28, 2000: Professor Patrick Cousot, École
Normale Supérieure, Paris.
July 29, 2000: Professor Richard Korf (AAAI Fellow),
University of California, Los Angeles.
An Overview of MAXQ Hierarchical Reinforcement Learning
Thomas G. Dietterich
Oregon State University, Corvallis, Oregon, USA
tgd@cs.orst.edu
Reinforcement learning addresses the problem of learning optimal
policies for sequential decision-making problems involving stochastic
operators and numerical reward functions rather than the more
traditional deterministic operators and logical goal predicates. In
many ways, reinforcement learning research is recapitulating the
development of classical research in planning and problem solving.
After studying the problem of solving ``flat'' problem spaces,
researchers have recently turned their attention to hierarchical
methods that incorporate subroutines and state abstractions. This
paper gives an overview of the MAXQ value function decomposition and
its support for state abstraction and action abstraction.
Partial Completeness of Abstract Fixpoint Checking
Patrick Cousot
Département d'informatique, École normale
supérieure, 45 rue d'Ulm
75230 Paris CEDEX 05, France
Patrick.Cousot@ens.fr
Abstract interpretation is used in program static analysis and model
checking to cope with infinite state spaces and/or with computer
resource limitations. One common problem is to check abstract
fixpoints for specifications. The abstraction is partially
complete when the checking algorithm is exact in that, if the
algorithm ever terminates, its answer is always affirmative for
correct specifications. We characterize partially complete
abstractions for various abstract fixpoint checking algorithms,
including new ones, and show that the computation of complete abstract
domains is essentially equivalent to invariance proofs that is to
concrete fixpoint checking.
Recent Progress in the Design and Analysisof Admissible
Heuristic Functions Richard E. Korf
Computer Science Department, University of California, Los Angeles
Los Angeles, CA 90095
korf@cs.ucla.edu
In the past several years, significant progress has been made in
finding optimal solutions to combinatorial problems. In particular,
random instances of both Rubik's Cube, with over $10^{19}$ states, and
the 5x5 sliding-tile puzzle, with almost $10^{25}$ states, have been
solved optimally. This progress is not the result of better search
algorithms, but more effective heuristic evaluation functions. In
addition, we have learned how to accurately predict the running time
of admissible heuristic search algorithms, as a function of the
solution depth and the heuristic evaluation function. One corollary
of this analysis is that an admissible heuristic function reduces the
effective depth of search, rather than the effective branching factor.
Copyright © 2000, American
Association of Artificial Intelligence (www.aaai.org). All rights
reserved.
|
![[SARA 2000]](sara2k-logo.gif){kind=logo}
