HOME   ::  Conference List   ::   Conference Paper
Fyfe, C. and Livingstone, D. (1997) Developing a Community Language. In ECAL97. Brighton, UK.
Bookmark:  

Full-text
   URL: ftp://ftp.cogs.susx.ac.uk/pub/ecal97/online/F161.ps.gz
   Cached: PDF-161K    PS-170K    PS.gz-65K   
   SAVE AS an easy-to-recall long filename:
      Filename format: author--year--title   PDF-161K    PS-65K    :: About GZip'd PS
      Filename format: author--year--title--journal|proceedings|...--pages   PDF-161K    PS-65K   

Related links
   CiteSeer: http://citeseer.ist.psu.edu/54114.html
  Web search: Google Web Search   ::   Google Scholar
  Within this site: Cited by (4)    References (10)

Paper at a Glance 

Developing a Community Language
Colin Fyfe and Daniel Livingstone,
Department of Computing and Information Systems,
The University of Paisley.
email: fyfe0ci,livi­ci0@paisley.ac.uk
Abstract We describe simulations of a community of agents who live in an environment which has some structure that the agents can learn to identify and subsequently about which they learn to commu­ nicate. Each agent has two entirely separate arti­ ficial neural networks which learn to perform the two tasks: identification of the structure in the environment and communication to others about the environment's structure. We show that a com­ munity of agents with similar representation ca­ pabilities is most successful in generating a com­ mon language and further that a community with different representation capabilities will evolve so that all members have the same representation capabilities.
1 Introduction This paper will investigate issues involved in the origin of a new language within a community of users of the language. We describe simulations involving agents sit­ uated within a common environment. These agents each have two entirely separate artificial neural networks: 1. The first encodes the environment by finding those sources which are creating the environment. This is done in an unsupervised way which leads to each agent having an unique internal representation of the environment. We shall see that it is beneficial to the community to have all agents within the environment sharing a common capacity for representation though not necessarily sharing a common representation. 2. The second takes the output of the first neural net­ work and encodes this in a way that is agreed by the community of agents. This second encoding is the shared communication in the community of agents. No agent can investigate the internal representation of the environment held by any other agent. Each agent however exists within the shared environment and can perceive the output/communication of the ...
BibTex
@inproceedings{fyfe97developingA,
  author={C. Fyfe and D. Livingstone},
  title={Developing a Community Language},
  year={1997},
  month={July},
  address={Brighton, UK},
  booktitle={ECAL97},
  url={http://www.isrl.uiuc.edu/~amag/langev/paper/fyfe97developingA.html}
}

 HOME   ::  Conference List   ::   Conference Paper Comments to: junwang4 you-know-at gmail.com Last update: 2/2/08