Cara Reedy, RPI Graduate Student

Cara Reedy, RPI Graduate Student

This talk will present an artificial life model consisting of agents that have diploid genomes, which create evolving neural networks that let the agents interact with their environment and each other.   Evolving the neural networks allows networks to search for solutions in ways that are different from typical learning algorithms. Diploidy (two copies of the genome), seen in many real-life organisms, keeps the population more diverse than the standard haploid (one copy of the genome) approach seen in most genetic algorithms and artificial life, and is better at adapting to changing environments. The goal of developing this simulation is to see if complex social issues, such as social discrimination, can be studied from a simplified model of biology and social relations between agents.

 

 

Share|