| 
  • If you are citizen of an European Union member nation, you may not use this service unless you are at least 16 years old.

  • You already know Dokkio is an AI-powered assistant to organize & manage your digital files & messages. Very soon, Dokkio will support Outlook as well as One Drive. Check it out today!

View
 

Chris Watkins talk

Page history last edited by Inman Harvey 11 years, 11 months ago

 NAME talk at Alergic, Wed 24 Oct 2007, 4:30pm in ARUN-401

 

Chris Watkins

Dept of Computer Science, Royal Holloway

The Channel Capacity of Selective Breeding

 

This talk will consider some familiar genetic and evolutionary algorithms from a new point of view. Instead of viewing these algorithms as optimisation methods, I will consider them information-theoretically, as communication channels. After a brief introduction to Shannon's theory of communication, I will show how to view an evolutionary algorithm as a communication channel, and then show that different algorithms have very different channel capacities.

     Intuitively, the channel capacity of an evolutionary algorithm is a measure of how much information -- or how much complexity -- can be put into the "genomes" as a result of selection. This seems a basic limitative computational property, which may be relevant to choosing useful regimes to get evolutionary algorithms to work effectively.

     In biological evolution,  in each generation genetic information is degraded by mutation, but also in some sense restored by selection. Some natural basic questions are: How much information could be encoded in a genome as a result of selection? How complex could organisms conceivably become? Does the potential complexity depend on whether the organisms are sexual? How could information be encoded most efficiently, in the sense that the greatest amount could be encoded for the lowest intensity of selection? The theoretical approach I will present may have some bearing on these questions.

Comments (0)

You don't have permission to comment on this page.