TY - GEN
T1 - Countering poisonous inputs with memetic neuroevolution
AU - Togelius, Julian
AU - Schaul, Tom
AU - Schmidhuber, Jürgen
AU - Gomez, Faustino
PY - 2008
Y1 - 2008
N2 - Applied to certain problems, neuroevolution frequently gets stuck in local optima with very low fitness; in particular, this is true for some reinforcement learning problems where the input to the controller is a high-dimensional and/or ill-chosen state description. Evidently, some controller inputs are "poisonous", and their inclusion induce such local optima. Previously, we proposed the memetic climber, which evolves neural network topology and weights at different timescales, as a solution to this problem. In this paper, we further explore the memetic climber, and introduce its population-based counterpart: the memetic ES. We also explore which types of inputs are poisonous for two different reinforcement learning problems.
AB - Applied to certain problems, neuroevolution frequently gets stuck in local optima with very low fitness; in particular, this is true for some reinforcement learning problems where the input to the controller is a high-dimensional and/or ill-chosen state description. Evidently, some controller inputs are "poisonous", and their inclusion induce such local optima. Previously, we proposed the memetic climber, which evolves neural network topology and weights at different timescales, as a solution to this problem. In this paper, we further explore the memetic climber, and introduce its population-based counterpart: the memetic ES. We also explore which types of inputs are poisonous for two different reinforcement learning problems.
UR - http://www.scopus.com/inward/record.url?scp=56449113460&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=56449113460&partnerID=8YFLogxK
U2 - 10.1007/978-3-540-87700-4_61
DO - 10.1007/978-3-540-87700-4_61
M3 - Conference contribution
AN - SCOPUS:56449113460
SN - 3540876995
SN - 9783540876991
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 610
EP - 619
BT - Parallel Problem Solving from Nature - PPSN X - 10th International Conference, Proceedings
T2 - 10th International Conference on Parallel Problem Solving from Nature, PPSN X
Y2 - 13 September 2008 through 17 September 2008
ER -