Evolutionary Computation

To view the full lecture on video, click "Here".

• What The News
• General information
  • When & Where
  • Tickets
  • Travel
  • FAQ
  • Thanks to ...
  • Contact
  • Forum (old site)
• Camping
  • Map of WTH
  • Weather
  • Hotels
  • Facilities
  • Campsite locations
• Participate
  • Volunteers
  • Villages
  • Projects
• Conference program
  • Schedule Day 1
  • Schedule Day 2
  • Schedule Day 3
  • Schedule Day 4
  • Events
  • Speakers
  • iCal
• Other
  • Photos
  • Press Reviews

Abstract

Evolutionary Computation is a term that describes a variety of methodsinspired by evolution that can solve problems that might otherwise beintractable. The methods share a population-based approach, in whichpossible solutions are "evolved" by performing "genetic operations" suchas mutation and cross-over on some type of genomic representation.

Using genetic programming methods you actually grow or evolvecomputer programs to solve your problem. Instead of specifying howsomething is to be done, you specify a "fitness measure" that tellshow good of a solution you've got, and then let the system evolve apopulation of candidates. Believe it or not, this stuff actuallyworks!

Human equivalent (or superior) solutions to problems such as analogcircuit design, control theory, circuit layout, synthesis of antennas,and optical systems have been achieved over the past couple of yearswith genetic programs running on 100 node clusters. Given Moore'slaw, this stuff will be running on our desktops soon.

Speakers

• Eric Blossom

Schedule

Informations