Adaptive Behavior, 3 (2)
 ISAB Home |
 SAB '08 |
| New Login |
| Log In |
| Journal |
| Conferences |
 Members |
| News |
| Joining ISAB |
| ISAB Officers |
| Contact ISAB |
Adaptive Behavior
Volume 3, Number 2
Fall 1994
Table of Contents
Dave Cliff and Susi Ross
Adding Tempory Memory to ZCS
Adaptive Behavior, 3 (2), 101-150.
Frédéric Gruau
Automatic Definition of Modular Neural Networks
Adaptive Behavior, 3 (2), 151-183.
Chisato Numaoka
Phase Transitions in Instigated Collective Decision Making
Adaptive Behavior, 3 (2), 185-223.
Peter M. Tood
Unsettling the Centralized Mindset
Review of Turtles, Termites, and Traffic Jams: Explorations in Massively Parallel Microworlds, by Mitchel Resnick. Cambridge, MA: MIT Press/Bradford Books, 1994.Recents Books of Interest
Adding Tempory Memory to ZCS
By Dave Cliff and Susi Ross
Abstract
In a recent article, Wilson (1994) described a "zeroth-level" classifier systems (ZCS). ZCS employs a reinforcement learning technique comparable to Q-learning (Watkins, 1989). This article presents results from the first reconstruction of ZCS. Having replicated Wilson's results, we extend ZCS in a manner suggested by Wilson: The original formulation of ZCS has no memory mechanisms, but Wilson (1994b) suggested how internal "temporary memory" register could be added. We show results from adding one-bit and two-bit memory registers to ZCS. Our results demonstrate that ZCS can exploit memory facilities efficiently in non-Markow environments. We also show that the memoryless ZCS can converge on near-optimal stochastic solutions in non-markow environments.We then presents results from trials using ZCS in Markov environments that require increasingly long chains of actions before reward is received. Our results indicate that inaccurate overgeneral classifiers can interact with the classifier-generation mechanisms to cause catastrophic breakdowns in overall system performance. Basing classifier fitness on accuracy may alleviate this problem. We conclude that the memory mechanism in its current form is unlikely to scale well for situations requiring large amounts of temporary memory. Nevertheless, the ability to find stochastic solutions when there is insufficient memory might offset this problem somewhat .
Key Words
classifier systems; ZCS; memory; action chains; FLAVA
Automatic Definition of Modular Neural Networks
By Frédéric Gruau
Abstract
This article illustrates an artificial developmental system that is a computationally efficient technique for the automatic generation of complex artificial neural networks (ANNs). The artificial developmental system can develop a graph grammar into a modular ANN made of a combination of simpler subnetworks. A genetic algorithm is used to evolve coded grammars that generate ANNs for controlling six-legged robot locomotion. A mechanism for the automatic definition of neural subnetworks is incorporated. Using this mechanism, the genetic algorithm can automatically decompose a problem into subproblems, generate a subANN for solving the subproblem, and instantiate copies of this subANN to build a higher-level ANN that solves the problem. We report some simulation results showing that the same problem cannot be solved if the mechanism for automatic definition of subnetworks is suppressed. We support our argument with pictures that describe the steps of development, how ANN structures are evolved, and how the ANNs compute.Key Words
animats; cellular encoding; modularity; locomotion; automatic definition of neural subnetworks
Phase Transitions in Instigated Collective Decision Making
By Chisato Numaoka
Abstract
This article proposes a computational model for an emergent collective behavior that collectively changes strategy type, such as from attack to defense, as seen in any kind of battle. It describes the result of an experimental simulation with multiple autonomous robots based on the proposed model.Our model first defines payoff functions that create multiple equilibrium states, each of which corresponds to one strategy type. Subsequently, we attempt to model the dynamics that cause the robots to change their choice of strategy type collectively when a small number of robots change their chosen type. In these dynamics, we pay particular attention to how many robots, called instigators, are required to make all robots eventually change their strategy type. In addition, to make it easy for all robots to change their strategy type, we provide a mechanism by which the robots themselves reduce the utility of strategies.
Key Words
collective behavior; global communication; instigators; decision making; nonlinear dynamics
Pages 225-229
Unsettling the Centralized Mindset
Peter M. Tood
Review of Turtles, Termites, and Traffic Jams: Explorations in Massively Parallel Microworlds, by Mitchel Resnick. Cambridge, MA: MIT Press/Bradford Books, 1994.
Pages 231-234
Recents Books of Interest
back to TOC, back to top