CALL FOR PAPERS

Workshop on Autonomy Oriented Computation(AOC)

at the 5th International Conference on Autonomous Agents

Montreal, May 29, 2001

(http://robotics.comp.hkbu.edu.hk/~jiming/aoc01.html)
(http://isg.enme.ucalgary.ca/aoc01/)

WORKSHOP OVERVIEW

Two interesting areas in autonomous agents, namely (1) synthetic autonomy and (2) multiagent approaches to complex systems, are fast growing and converging. Some examples are ALIVE, Artificial Fish, Boids, SWARM, and ANT systems. Lifelike behavior and/or emergent intelligence have been exhibited in these systems by means of constructing and operating artifacts. Other research, such as Internet ecology, statistical mechanics, immune networks, and dynamic economies, has proposed approaches to understand self-organized phenomena by modeling and simulating autonomous entities.

While existing approaches to modeling autonomy are successful to some extent, a generic model or architecture to solve problems in such complex systems effectively is still absent. A new and promising concept, namely Autonomy-Oriented Computation (AOC), is needed to unify the methodologies for effective analysis, modeling, and simulation of the characteristics of such complex systems as ecological systems, social systems, biological systems, economical systems, physical and chemical systems, and natural systems. AOC is an attempt to provide a new computational paradigm that makes use of the autonomous nature of individual entities in complex systems. Comparing to other paradigms, such as multiagent-based design/modeling, artificial life, and evolutionary algorithms, the abilities of AOC will be appealing. The intent of this workshop is to highlight and start addressing the theoretical and practical issues concerning AOC.

MAJOR OBJECTIVES OF AOC

• To capture the abstract essence of an autonomous system.

• To solve the problem of computational intractability when applying conventional models, e.g., high-dimensionality. A typical example would be to study population behavior in ecology, biology, economics, or social science, when the number of population genomes is very large and the traditional mathematical modeling becomes intractable.

• To observe and study complex phenomena or behaviors from a system that involves many competing bodies as well as layers of interactions among them. The complexity of such systems CANNOT be reduced to the basic/fundamental laws of phenomena or behaviors that we are often familiar with, such as physics. For instance, a stock price may grow, that globally follows one (e.g., exponential) distribution and locally follows another (e.g., logistic). Also, many systems may behave very chaotic at a microscopic level, but can achieve overall equilibrium easily.

• To observe and study: (1) emergent behavior (not evident in rules) - in many cases the apparent autonomy or complexity of living organisms can be recreated from some rigid rules, (2) collective behavior (not the characteristics or properties of individual agents), and (3) spontaneous order.

• To observe and study complex phenomena or behaviors. The complexity of such systems may be created through an iterative process that is regulated via a set of rules, such as using L-systems to create a tree. The rules can get very complex. For instance, in the case of L-systems, self-avoiding or randomized or biased parameterization may be added. It is no surprise now that some simple rule-based system can achieve computational ability of a programmable computer.

• To study the complexity of a certain problem by examining its possible diversity (such as the famous Volterra's principle in the case of many species; we are interested in examining where or when steady states, periodic cycles, or chaos/bifurcation will occur).

• To understand or discover the underlying mechanisms or origins of complex phenomena or behaviors. The possible strategies of mutated HIV virus in attaching a host, apart from its high turnover rate -- this result can have implications to developing new drugs.

• To observe, change, examine long-term behaviors.

TOPICS

Topics of the workshop include, but are not limited to, the following areas:

1. Methodology, Theory and Perspectives of AOC. Measurement of emergence; measurement of evolvability; self-organization in complex systems; behavioural monitoring of autonomous societies; performance measurement for AOC-based systems; formation of roles and social structure in the communities; embodiment of autonomous entities; and dialectics of microscopic and macroscopic autonomies

2. Implementation Issues. Guidelines for designing AOC; simulating environments and languages for AOC; architectural issues; tractability and scalability of algorithms; visualization of activities in the testing environments; and the design of local and global interaction rules.

3. Applications. Examples of successful application of AOC to real-life problems; potential application areas of AOC (e.g. distributed search, financial market modelling, data analysis).

4. Comparisons. Strength and weaknesses of AOC vs. other multiagent paradigms such as evolutionary computation, multiagent simulation, emergent computation, artificial life, L-systems, evolutionary strategies, cellular automata; and empirical performance comparison using benchmark problems.

PAPER SUBMISSION

Papers should report original work and should not exceed 10 pages including all figures in the same format as the main conference proceedings. All papers will be reviewed by the programme committee, and selected on their originality, timeliness, relevance and clarity.

Electronic submission is preferred. Please email a PostScript or PDF copy of your submission to KC Tsui (tsuikc@comp.hkbu.edu.hk) before March 9, 2001. You may also send hard copies to

Dr. K C Tsui
Department of Computer Science
Hong Kong Baptist University
224, Waterloo Rd, Kowloon Tong
Hong Kong.

IMPORTANT DATES

PAPER PRESENTATION

All presentations must be between 20 to 25 minutes. This will be followed by a directed discussion.

The workshop will be concluded by a panel discussion on the main topics covered by and issues rising from the presentations.

PUBLICATIONS

Selected workshop papers will be published as a Special Issue in an international journal, and/or as an edited volume by an international publisher.

TENTATIVE PROGRAM

1:30pm Welcome

1:30-3:15pm Session 1

1. Introducing Autonomy Oriented Computation (Jiming Liu, Kwok Ching Tsui and Jianbing Wu)

2. Analysis of Adaptive Decision-Making Frameworks (K. S. Barber, I. M. Gamba and C. E. Martin)

3. What Kinds of Properties Determine Characteristics of Multiple Learning Agents? - Implications from goal and evaluation in agents (Keiki Takadama and Katsunori Shimohara)

4. Reflexivity and Meta-reasoning in Multi-Agent Systems (Michal Pechoucek, Douglas Norrie and Vladimir Marik)

3:15-3:45pm Break

3:45-5:30pm Session 2

1. Multi-Agent Systems as Social Autonomous Systems (Sigmar Papendick, Jorg Wellner and Werner Dilger)

2. Fundamental Issues in the Use of Genetic Programming in Agent Based Computational Economics (Shu-Heng Chen)

3. Mechanisms and Military Applications for Synthetic Pheromones (H. Van Dyke Parunak, Sven Brueckner, John Sauter and Jeff Posdamer)

4. Autonomy-Oriented Computation in Pheromone Robotics (David Payton, Mike Daily, Bruce Hoff, Mike Howard and Craig Lee

5:30-6:00pm Panel Discussion

WORKSHOP ORGANIZERS

PROGRAM COMMITTEE