collins@z.ils.nwu.edu (Gregg Collins) (06/22/91)
What follows is the schedule for the Eighth International Workshop on
Machine Learning, which will be held in Evanston, Illinois on the
27th through 29th of this month. For those who have registered: If
you have not received confirmation of your registration, please send
email to collins@ils.nwu.edu, or (if necessary) call (708) 491-7340.
********************************************************************
ML91 MASTER SCHEDULE
====================
THURSDAY, JUNE 27
8:15-9:00 Continental breakfast
9:00-10:00 ML91 invited address
Jaime Carbonell, CMU
10:00-10:30 Break
10:30-12:00 Workshop sessions
12:00-1:30 Lunch
1:30-3:00 Workshop sessions
3:00-3:30 Break
3:30-5:30 Workshop sessions
6:00-9:00 ML91 poster session
FRIDAY, JUNE 28
8:15-9:00 Continental breakfast
9:00-10:30 Workshop sessions
10:30-11:00 Break
11:00-12:30 Workshop sessions
12:30-2:00 Lunch
2:00-3:00 ML91 invited address
Doug Medin, U. of Michigan
3:30-5:00 Workshop sessions
5:30-7:00 Cocktail party
SATURDAY, JUNE 29
8:15-9:00 Continental breakfast
9:00-10:30 Workshop sessions
10:30-11:00 Break
11:00-12:00 ML91 invited address
Judea Pearl, UCLA
12:00-1:30 Lunch/ ML business meeting
1:30-3:00 Workshop sessions
3:00-3:30 Break
3:30-5:00 ML91 summary panel session
************************************************************************
AUTOMATED KNOWLEDGE ACQUISITION
===============================
THURSDAY, JUNE 27
9:00-10:00 ML91 invited address
Jaime Carbonell, CMU
6:00-9:00 ML91 poster session
FRIDAY, JUNE 28
9:00-10:30 Paper Presentations (David Wilkins, chair)
o Improving the Performance of Inconsistent Knowledge Bases via Combined
Optimization Method
Y. Ma and D. Wilkins, U. of Illinois
o Knowledge Acquisition Combining Analytical and Empirical Techniques
M. Martin, R. Sanguesa, and C. Ulises
o A Domain-Independent Framework for Effective Experimentation in
Planning
Y. Gil, CMU
11:00-12:30 Panel (Tom Gruber, chair)
Capturing Design Rationale
2:00-3:00 ML91 invited address
Doug Medin, University of Michigan
3:30-5:00 Paper presentation (Ray Bareiss, chair)
o Generating Error Candidates for Assigning Blame in a Knowledge Base
M. Weintraub and T. Bylander, Ohio State U.
o The Flexibility of Speculative Refinement
S. Craw and D. Sleeman, King's College
o Knowledge Refinement Using a High-Level, Non-Technical Vocabulary
E. Jones, Northwestern U.
5:30-7:00 Cocktail party
SATURDAY, JUNE 29
9:00-10:30 Workshop invited address
Bruce Buchanan, University of Pittsburg
11:00-12:00 ML91 invited address
Judea Pearl, UCLA
1:30-3:00 Working group sessions
(Topics are to be determined by workshop participants)
3:30-5:00 ML91 summary panel session
****************************************************************************
CONSTRUCTIVE INDUCTION
======================
THURSDAY, JUNE 27
9:00-10:00 ML91 invited address
Jaime Carbonell, CMU
10:30-12:00 Theoretical issues
o The need for constructive induction
C. Matheus, GTE Laboratories
o On the effect of instance representation on generalization
S. Saxena, U. of Massachusetts
o Quantifying the value of constructive induction, knowledge, and noise
filtering on inductive learning
C. Kadie, U. of Illinois
1:30-3:00 Empirical symbolic approaches
o ID2-of-3: Constructive induction of m-of-n concepts for discriminators
in decision trees
P. Murphy and M. Pazzani, UC Irvine
o Constructive induction on symbolic features: Introducing new
comparative terms
B. Leng and B. Buchanan, U. of Pittsburgh
o A scheme for feature construction and a comparison of empirical methods
D. Yang, L. Rendell, and G. Blix, U. of Illinois
3:30-5:30 Knowledge-based approaches I
o Relations, domain knowledge and disjunctive concept learning
H. Ragavan and L. Rendell, U. of Illinois
o Constructive induction on domain information
J. Callan and P. Utgoff, U. of Massachusetts
o Opportunistic constructive induction: Using fragments of domain
knowledge to guide construction
G. Gunsch, Air Force Institute of Technology
o A hybrid method for feature generation
T. Fawcett and P. Utgoff, U. of Massachusetts
6:00-9:00 ML91 poster session
FRIDAY, JUNE 28
9:00-10:30 Knowledge-based approaches II
o Constructive induction in knowledge-based neural networks
G. Towell, M. Craven, and J. Shavlik, U. of Wisconsin
o Learning variable descriptors for applying heuristics across CSP
problems
D. Day, MITRE
o Informed pruning in constructive induction
G. Drastal, Siemens Corporate Research
11:00-12:30 Relational/structural approaches
o Learning structural decision trees from examples
L. Watanabe and L. Rendell, U. of Illinois
o Relational cliches: Constraining constructive induction during
relational learning
G. Silverstein and M. Pazzani, UC Irvine
o Constraints on predicate invention
R. Wirth and P. O'Rorke, UC Irvine
2:00-3:00 ML91 invited address
Doug Medin, U. of Michigan
3:30-5:00 Connectionist approaches
o A neural network approach to constructive induction
D. Yeung, U. of Hong Kong
o Learning polynomial functions by feature construction
R. Sutton and C. Matheus, GTE Laboratories
o Discovering production rules with higher order neural networks: A
case study
A. Kowalczyk, Telecom Australia Research Laboratory
o Learning concepts by synthesizing minimal threshold gate networks
A. Oliveira and A. Sangiovanni-Vincentelli, UC Berkeley
5:30-7:00 Cocktail party
SATURDAY, JUNE 29
9:00-10:30 Incremental learning, theory refinement and discovery
o Incremental constructive induction: An instance-based approach
D. Aha, Turing Institute
o Constructive induction in theory refinement
R. Mooney and D. Ourston, U. of Texas
o Constructive induction of numerical terms in automated discovery of
empirical equations
R. Zembowicz and J. Zytkow, Wichita State U.
11:00-12:00 ML91 invited address
Judea Pearl, UCLA
1:30-2:30 Inverse resolution
o Abstracting concepts with inverse resolution
A. Giordana, D. Roverso, and L. Saitta, U. of Turin
o A critical comparison of various methods based on inverse resolution
X. Ling and M. Narayan, U. of Western Ontario
2:30-3:00 Workshop panel discussion
3:30-5:00 ML91 summary panel session
************************************************************************
LEARNING FROM THEORY AND DATA
=============================
THURSDAY, JUNE 27
9:00-10:00 ML91 invited address
Jaime Carbonell, CMU
10:30-12:00 Introductions to "Learning from Theory and Data"
o Introduction to this Workshop
o The Generality of Overgenerality
W. Cohen, AT&T Bell Laboratories
o Is it a Pocket or a Purse: Tightly Coupled Theory and Data
Driven Learning
E. Wisniewski and D. Medin, U. of Michigan
1:30-3:00 Empirical Theory Revision 1
o Improving Shared Rules in Multiple Category Domain Theories
D. Ourston and R. Mooney, U. of Texas
o The DUCTOR: A Theory Revision System for Propositional Domains
T. Cain, UC Irvine
o A Study of How Domain Knowledge Improves Knowledge-Based Learning
Systems
B. Whitehall and S. Lu, U. of Illinois
3:30-5:30 Empirical Theory Revision 2
o Revision Cost for Theory Refinement
R. Hamakawa, CC&C Systems Research Laboratories, NEC
o Incremental Refinement of Approximate Domain Theories
R. Feldman and A. Segre, Cornell
o Revision of Reduced Theories
Z. Ling and M. Valtorta, U. of Western Ontario
o Poster previews
6:00-9:00 ML91 poster session
FRIDAY, JUNE 28
9:00-10:30 Revision of Non-Standard Domain Theories
o Refining Domain Theories Expressed as Finite-State Automata
R. Maclin and J. Shavlik, U. of Wisconsin
o Refinement of Approximate Reasoning-Based Controllers by Reinforcement
Learning
H. Berenji, NASA Ames
o A Smallest Generalization Step Strategy to Deal with Over Generalization
and Complete Domain Theory
C. Nedellec, Universite Paris-Sud
11:00-12:30 Paper presentations
o Probabilistic Evaluation of Bias for Learning Systems
M. desJardins, Berkeley
o Using a Domain Theory to Improve Learning Rate in Concept Formation
K. Thompson and W. Iba, NASA Ames
o Identifying Cost Effective Boundaries of Operationality
J. Yoo and D. Fisher, Vanderbilt
2:00-3:00 ML91 invited address
Doug Medin, U. of Michigan
3:30-5:00 Group discussions
o Standardizing Terminology for Integrated Learning Algorithms and
Imperfect Theories
o Creating Shared Testbeds for Empirical Studies of Integrated Learning
5:30-7:00 Cocktail party
SATURDAY, JUNE 29
9:00-10:30 Learning with Multiple Reasoning Methods
o A Method for Multistrategy Task-Adaptive Learning based on
Plausible Justifications
G. Tecuci and R. Michalski, George Mason
o Improving Learning Using Causality and Abduction
M. Botta, S. Ravotto, L. Saitta, and S. Sperotto, U. di Torino
o Discovering Regularities from Large Knowledge Bases
W. Shen, Microelectronics and Computer Technology Corp.
11:00-12:00 ML91 invited address
Judea Pearl, UCLA
1:30-3:00 Integrating Learning and Problem Solving
o An Enhancer for Reactive Plans
D. Gordon, Naval Research Laboratory
o A Hybrid Approach to Guaranteed Effective Control Strategies
J. Gratch and G. DeJong, U. of Illinois
o Learning with Inscrutable Theories
P. Tadepalli, Oregon State U.
3:30-5:00 ML91 summary panel session
*************************************************************************
LEARNING RELATIONS
==================
THURSDAY, JUNE 27
9:00-10:00 ML91 invited address: Jaime Carbonell, CMU
10:30-12:00 Constraining first-order learning
o Learning Constrained Atoms
C.D. Page, A. Frisch, U. of Illinois
o Integrity Constraints and Interactive Concept-Learning
L. de Raedt, M. Bruynooghe, B. Martens
1:30-3:00 Information-based Approaches to First Order Learning
o Determinate Literals as an Aid in Inductive Logic Programming
J. R. Quinlan, GTE Laboratories
o Learning relations from noisy examples: An empirical comparison of
LINUS and FOIL
S. Dzeroski and N. Lavrac, Jozef Stefan Institute
o Noise-Tolerant Relational Concept Learning Algorithms
C. Brunk, M. Pazzani, UC Irvine
3:30-5:30 Logical generalization
o Learning Qualitative Models of Dynamic Systems
I. Bratko, S. Muggleton, A. Varsek
o Inducing Temporal Fault Diagnostic Rules from a Qualitative Model
C. Feng
o Experiments in Non-Monotonic First-Order Induction
M. Bain, Turing Institute
6:00-9:00 ML91 poster session
FRIDAY, JUNE 28
9:00-10:30 Induction in Relational Relational Learning
o Efficient Learning of Logic Programs with Non-determinate,
Non-discriminating Literals
B. Kijsirikul, M. Numao, M. Shimura, Tokyo Institute of Technology
o Learning Search Control Rules for Planning: an Inductive Approach
C. Leckie, I. Zukerman
o The Consistent Concept Axiom
Z. Qian, K. Irani, U. of Michigan
11:00-12:30 Exploiting Prior Knowledge
o A Knowledge-Intensive Approach to Learning Relational Concepts
M. Pazzani, C. Brunk, G. Silverstein, UC Irvine
o First-Order Theory Revision
B. Richards, R. Mooney, U. of Texas
o Revising Relational Domain Theories
J. Wogulis, UC Irvine
2:00-3:00 ML91 invited address
Doug Medin, U. of Michigan
3:30-5:00 Inverting Resolution
o Using Inverse Resolution to Learn Relations from Experiments
D. Hume, C. Sammut
o Constraints on Predicate Invention
R. Wirth, P. O'Rorke, UC Irvine
o Completeness for Inductive Procedures
C. Rouveirol, Katholieke Universitat Leuven
5:30-7:00 Cocktail party
SATURDAY, JUNE 29
9:00-10:30 Unsupervised approaches
o Learning Stochastic Motifs from Genetic Sequences
K. Yamanishi, A. Konogaya, NEC Corporation
o COBWEB-R: Concept Formation with Relational Data
J. Allen, K. Thompson NASA Ames
o Learning Spatial Relations from Images
K. Hiraki, J. Gennari, Y. Yamamoto, Y. Anzai, Keio U.
11:00-12:00 ML91 invited address
Judea Pearl, UCLA
1:30-3:00 Wrap-up
Discussion of Relational Learning papers
3:30-5:00 ML91 summary panel session
**************************************************************************
ENGINEERING AUTOMATION
======================
THURSDAY, JUNE 27
9:00-10:00 ML91 invited address: Jaime Carbonell, CMU
10:30-12:00 Machine Learning and Large Engineering Datasets
o Conceptual Clustering and Exploratory Data Analysis
G. Biswas, J. Weinberg, D. Yang and G. Koller, Vanderbilt U.
o Megainduction: a test flight
J. Catlett, U. of Sydney
1:30-3:00 Machine Learning and Engineering Modelling
o Knowledge-based Equation Discovery for Engineering Problems
B. Rao, S. Lu, B. Stepp, U. of Illinois
o AIMS: An Adaptive Interactive Modelling System
D. Tcheng, B. Lambert, S. Lu and L. Rendell, U. of Illinois
o Model Revision: A Theory of Incremental Model Learning
A. Goel, Georgia Tech
3:30-5:30 Panel discussion
Machine Learning and Engineering Modelling
6:00-9:00 ML91 poster session
FRIDAY, JUNE 28
9:00-10:30 Machine Learning in Robotics and Maufacturing Automation
o Continuous Conceptual Set Covering: Learning Robot Operators
from Examples
C. Kadie, U. of Illinois
o Comparing Stochastic Planning to the Acquisition of Increasingly
Permissive Plans for Complex, Uncertain Domains
S. Bennett, G. DeJong, U. of Illinois
o Decision Tree Induction of 3-D Manufacturing Features
L. Watanabe, S. Yerramareddy, U. of Illinois
11:00-12:30 Panel Discussion
Machine Learning for Engineering Automation: An Engineer's Wish List
2:00-3:00 ML91 invited address
Doug Medin, U. of Michigan
3:30-5:00 Machine Learning for Engineering Design
o Learning Analytical Knowledge about VLSI Design from Observation
J. Herrman, Universitat Dortmund
o Self-modelling databases
J. Schlimmer, Carnegie-Mellon U.
o Knowledge Compilation to Speedup Numerical Optimization Tasks
G. Cerbone, T. Dietterich, Oregon State U.
o Designing Integrated Learning Systems for Engineering Design
Y. Reich, Carnegie-Mellon U.
5:30-7:00 Cocktail party
SATURDAY, JUNE 29
9:00-10:30 Machine Learning and Noise/Preprocessing
o Improving Recognition Effectiveness of Noisy Texture Concepts
through Optimization of Their Descriptions
P. Pachowicz and J. Bala, George Mason U.
o Noise Resistant Classification: Subsymbolic and Hybrid Architectures
for Event Classification in Plasma Physics
L. Belyaev and L. Falcone, JPL
o Machine Learning for Nondestructive Evaluation
P. O'Rorke, S. Morris, Amirfathi, Bond and D. St. Clair, UC Irvine
and McDonnell Douglas
11:00-12:00 ML91 invited address
Judea Pearl, UCLA
1:30-3:00 Workshop session
Focus groups
3:30-5:00 ML91 summary panel session
*************************************************************************
LEARNING REACTION STRATEGIES
============================
THURSDAY, JUNE 27
9:00-10:00 ML91 invited address: Jaime Carbonell, CMU
10:30-12:00 Learning World Models (Charles Martin, chair)
o Learning to Select a Model in a Changing World
M. Kokar, Northeastern U.
o Learning a Set of Primitive Actions with an Uninterpreted
Sensorimotor Apparatus
D. Pierce, U. of Texas
o The Blind Leading the Blind: Mutual Refinement of Approximate Theories
S. Kedar, NASA Ames
1:30-3:00 Deliberation and Reaction (Reid Simmons, chair)
o Becoming Decreasingly Reactive: Learning to Deliberate Minimally
S. Chien, JPL
o Decision-Theoretic Learning in an Action System
M. Brand, Northwestern U.
o Learning from Deliberated Reactivity
B. Krulwich, Northwestern U.
3:30-5:30 Workshop invited speaker
Chris Atkeson, MIT
6:00-9:00 ML91 poster session
FRIDAY, JUNE 28
9:30-10-30 Workshop invited speaker
Andy Barto, U. of Massachusetts
11:00-12:30 Reinforcement Learning and Dynamic Programming
(Richard Sutton GTE, chair)
o Self-Improving Based on Reinforcement Learning
L. Lin, CMU
o Variable Resolution Dynamic Programming: Efficiently Learning Action
Maps in Multivariate Real-Valued State Spaces
A. Moore, MIT
o Planning by Incremental Dynamic Programming
R. Sutton, GTE Research Labs
2:00-3:00 ML91 invited address
Doug Medin, U. of Michigan
3:30-5:00 Learning in Robotics (Leslie Kaelbling, chair)
o Learning to Avoid Obstacles through Reinforcement
J.R. Millan
o Scaling Reinforcement Learning to Robotics by Exploiting the
Subsumption Architecture
S. Mahavaden, IBM T.J. Watson Research Center
o Learning Footfall Evaluation for a Walking Robot
G. Hsu, CMU
5:30-7:00 Cocktail party
SATURDAY, JUNE 29
9:00-10:30 Learning Complex Operators (Jim Firby, chair)
o Learning a Cost-Sensitive Internal Representation for Reinforcement
Learning
M. Tan, CMU
o Learning the Persistence of Actions in Reactive Control Rules
H. Cobb
o Complexity and Cooperation in Q-Learning
S. Whitehead, Rochester
11:00-12:00 ML91 invited address
Judea Pearl, UCLA
1:30-3:00 Modular Learning Systems (Whitehead, Chair)
o Incremental Development of Complex Behaviors through Automatic
Construction of Sensory-motor Heirarchies
M. Ring, U. of Texas
o Scaling Reinforcement Learning Techniques via Modularity
L. Wixson, Teleos
o Transfer of Learning Across Compositions of Sequential Tasks
S.P. Singh, U. of Massachusetts
3:30-5:00 ML91 summary panel session
************************************************************************
COMPUTATION MODELS OF HUMAN LEARNING
====================================
THURSDAY, JUNE 27
9:00-10:00 ML91 invited address: Jaime Carbonell, CMU
10:30-12:00 Language
o A psychologically plausible cross-linguistic model of lexical and
syntactic acquisition
R. Kazman, CMU
o Language learning and psychological predictions: Computer modelling
of acquisition orders in child language
S. Nicholl and D. Wilkins, U. of Illinois
o Learning words from context
P. Hastings S. Lytinen, and R. Lindsay. U. of Michigan
1:30-3:00 Categories and Structures
o A constraint-motivated lexical acquisition model
C. Miller and J. Laird, U. of Michigan
o Variability bias and category learning
J. Martin D. Billman, Georgia Tech
o Simulating stages of human cognitive development with connectionist
models
T. Shultz, McGill U.
3:30-4:30 Categories
o Combining evidence of deep and surface similarity
D. Fisher J. Yoo, Vanderbilt
o A prototype based symbolic concept learning system
M. de la Maza
4:30-5:30 Mini-Symposium: Reorganization and Representation Change
in Learning Systems
D. Billman, Georgia Tech
T. Shultz, McGill U.
J. Pollack, Ohio State
6:00-9:00 ML91 poster session
FRIDAY, JUNE 28
9:00-10:30 Senory-Motor/Perception-Action
o Maeander: A model of human motor behavior and learning
W. Iba, NASA Ames
o Adaptive pattern-oriented chess
R. Levinson and R. Snyder, UC Santa Cruz
o Internal world models and supervised learning
M. Jordan and D. Rumelhart, Stanford
11:00-12:30 Skills
o A computational model of acquisition for children's addition strategies
R. Jones, LRDC
o The acquisition of expertise in human planning
P. Langley and J. Allen, NASA Ames
o Learning physics via explanation-based learning of correctness and
analogical search control
K. VanLehn and R. Jones, U. of Pittsburg
2:00-3:00 ML91 invited address
Doug Medin, U. of Michigan
3:30-5:00 Analogy/Skills
o The importance of causal structure and facts in evaluating explanations
M. Gick and S. Matwin
4:00-5:00 Mini-Symposium: On the relation between psychology and
machine learning
P. Langley, NASA Ames
R. Sutton, GTE Research Labs
C. Siefert, U. of Michigan
K. VanLehn, U. of Pittsburg
5:30-7:00 Cocktail party
SATURDAY, JUNE 29
11:00-12:00 ML91 invited address
Judea Pearl, UCLA
3:30-5:00 ML91 summary panel session
**************************************************************************
MACHINE LEARNING AND INTELLIGENT INFORMATION RETRIEVAL
======================================================
THURSDAY, JUNE 27
9:00-10:00 ML91 invited address: Jaime Carbonell, CMU
10:30-10:45 Intro, define working groups
10:45-12:05 Paper presentations
o A goal-based approach to intelligent IR
A. Ram, Georgia Tech, and L. Hunter, NIH
o Machine Learning in the Combination of Expert Opinion approach to IR
P. Thompson, PRC Incorporated
o Query Formulation through knowledge acquisition
J. Deogun, S. Bhatia, and V. Raghavan
o Classification trees for IR
S. Crawford
12:05-1:20 Lunch with working group members
1:20-3:00 Working groups meet
3:30-5:10 Paper presentations
o Query learning in information retrieval using an artificial neural
network with adaptive architecture
K. Kwok, Queens College, CUNY
o Incremental learning in a probabilistic information retrieval system
A. Goker and T. McCluskey, City University
o A probabilistic retrieval scheme for cluster-based adaptive information
retrieval
J. Bhuyan and V. Raghavan
o Conceptual clustering and exploratory data analysis
G. Biswas, J. Weinberg, Q. Yang, G. Koller, Vanderbilt U.
o Predicting actions from induction on past performance
S. Walczak, U. of Florida
5:10-5:30 Working group reports
6:00-9:00 ML91 poster session
FRIDAY, JUNE 28
2:00-3:00 ML91 invited address
Doug Medin, U. of Michigan
5:30-7:00 Cocktail party
SATURDAY, JUNE 29
11:00-12:00 ML91 invited address
Judea Pearl, UCLA
3:30-5:00 ML91 summary panel session