Specialization in multi-agent systems through learning.

A Murciano; J R Millán; J Zamora

doi:10.1007/s004220050351

Home/Archive/PubMed

PubMed · 9237363

Specialization in multi-agent systems through learning.

A Murciano·J R Millán·J Zamora

View original source ↗Export metadata ↓

Abstract

Specialization is a common feature in animal societies that leads to an improvement in the fitness of the team members and to an increase in the resources obtained by the team. In this paper we propose a simple reinforcement learning approach to specialization in an artificial multi-agent system. The system is composed of homogeneous and non-communicating agents. Because there is no communication, the number of agents in the team can easily scale up. Agents have the same initial functionalities, but they learn to specialize and so cooperate to achieve a complex gathering task efficiently. Simulation experiments show how the multi-agent system specializes appropriately so as to reach optimal (or near-to-optimal) performance in unknown and changing environments.

Explore related subjects

Animal Communication Animals Behavior, Animal Environment Design Learning Models, Biological Journal Article

Keep this discovery

Explore connections, maps & timelines

BibTeX RIS

A Murciano, J R Millán, J Zamora. 1997. Specialization in multi-agent systems through learning.. https://doi.org/10.1007/s004220050351

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related citations

PubMed1999

Communication goes multimodal.

Explore the source record for details and available documents.

Animal Communication↗

PubMed1999

Design features for electric communication.

How do the communication discharges produced by electric fish evolve to accommodate the unique design features for the modality? Two design features are considered: first, the limited range of signaling imposed on the electric modality by the physics of signal transmission from dipole sources; and second, the absence of signal echoes and reverberations for electric discharges, which are non-propagating electrostatic fields. Electrostatic theory predicts that electric discharges from fish will have a short range because of the inverse cube law of geometric spreading around an electrostatic dipole. From this, one predicts that the costs of signaling will be high when fish attempt to signal over a large distance. Electric fish may economize in signal production whenever possible. For example, some gymnotiform fish appear to be impedance-matched to the resistivity of the water; others modulate the amplitude of their discharge seasonally and diurnally. The fact that electric signals do not propagate, but exist as electrostatic fields, means that, unlike sound signals, electric organ discharges produce no echoes or reverberations. Because temporal information is preserved during signal transmission, receivers may pay close attention to the temporal details of electric signals. As a consequence, electric organs have evolved with mechanisms for controlling the fine structure of electric discharge waveforms.

Animal Communication↗

PubMed1999

Neural circuitry for communication and jamming avoidance in gymnotiform electric fish.

Over the past decade, research on the neural basis of communication and jamming avoidance in gymnotiform electric fish has concentrated on comparative studies of the premotor control of these behaviors, on the sensory processing of communication signals and on their control through the endocrine system, and tackled the question of the degree to which these behaviors share neural elements in the sensory-motor command chain by which they are controlled. From this wealth of investigations, we learned, first, how several segregated premotor pathways controlling a single central pattern generator, the medullary pacemaker nucleus, can provide a large repertoire of behaviorally relevant motor patterns. The results suggest that even small evolutionary modifications in the premotor circuitry can yield extensive changes in the behavioral output. Second, we have gained some insight into the concerted action of the brainstem, the diencephalon and the long-neglected forebrain in sensory processing and premotor control of communication behavior. Finally, these studies shed some light on the behavioral significance of multiple sensory brain maps in the electrosensory lateral line lobe that long have been a mystery. From these latter findings, it is tempting to interpret the information processing in the electrosensory system as a first step in the evolution towards the 'distributed hierarchical' organization commonly realized in sensory systems of higher vertebrates.

Animal Communication↗