2000 News___________________________________________

August 2000

JRI is pleased to announce that the paper "Levels of Compartmentalization in Artifical Life" has been accepted to the Artifical Life 7 Conference to be held on August 1st through August 6th. People who worked on this paper include, Research Mentor Dr. Sanza Kazadi, Daniel Lee, Rusha Modi, Jeff Sy, and Waynn Lue. Research on this type of artificial evolution is an ongoing research project that has been in development for over three years and will foreseeably go on for several more.

 

Class of 2000

June 2000

JRI is proud to present the graduating class of 2000. This year, JRI has five graduates. Their level of achievement is outstanding and a credit to their hard work, their families, and to JRI.

• Jey Chung will be attending Boston University Medical Program this coming school year.
  
• Daniel Lee will be attending the Massachusettes Institute of Technology.
  
• Rusha Modi will be attending the Brown University Medical Program.
  
• Sarah Rhee will be attending the Massachusettes Institute of Technology.
  
• Jeff Sy will be attending the University of Southern California.

March 2000

A new research group has been formed at JRI.  This research group is investigating the use of generic modules in the design of robotic systems. The group is made up of six individuals, five of whom are from Caltech, and three of which are undergraduates.

March 2000

JRI is pleased to report that a new paper entitled " Levels of Compartmentalization in Artificial Evolution " has been accepted for publication and presentation in the GECCO 2000 conference.  This work was completed by students Daniel Lee, Rusha Modi, Jeff Sy, and Waynn Lue under the guidance of Research Mentor Sanza Kazadi.  The paper addresses a mathematical limitation in many of the encodings currently employed in evolutionary design programs.  Three different types of encodings are possible.  One is  a noncompartmentalized encoding, in which the encoding has no built in structure, and each element is free to change independently.  The second type is a partially compartmentalized encoding.  This is an encoding in which the DNA has some structure, but this structure has no inherant "physics" behind it to form the structure into a complete design.  The final type is a fully compartmentalized model in which all functions are performed in a discrete compartment, and the way in which the compartments connect up is both part of the model, and part of the evolution.  Examples are worked indicating that the potential gains are extraordinary.