Abstract | ||
---|---|---|
In this paper, we investigate the use of nested evolution in which each step of one evolutionary process involves running a second evolutionary process. We apply this approach to build an evolutionary system for reinforcement learning (RL) problems. Genetic programming based on a descriptive encoding is used to evolve the neural architecture, while an evolution strategy is used to evolve the connection weights. We test this method on a non-Markovian RL problem involving an autonomous foraging agent, finding that the evolved networks significantly outperform a rule-based agent serving as a control. We also demonstrate that nested evolution, partitioning into subpopulations, and crossover operations all act synergistically in improving performance in this context. |
Year | DOI | Venue |
---|---|---|
2009 | 10.1145/1569901.1570034 | GECCO |
Keywords | Field | DocType |
evolutionary process,connection weight,non-markovian reinforcement learning,non-markovian rl problem,nested evolution,evolution strategy,autonomous agent,rule-based agent,autonomous foraging agent,evolutionary system,crossover operation,act synergistically,genetic programming,rule based,reinforcement learning | Evolutionary acquisition of neural topologies,Autonomous agent,Crossover,Evolutionary robotics,Computer science,Genetic programming,Evolution strategy,Artificial intelligence,Evolutionary programming,Machine learning,Reinforcement learning | Conference |
Citations | PageRank | References |
1 | 0.38 | 16 |
Authors | ||
2 |
Name | Order | Citations | PageRank |
---|---|---|---|
Jae-Yoon Jung | 1 | 297 | 31.94 |
James A. Reggia | 2 | 1000 | 276.13 |