EANT+KALMAN: An Efficient Reinforcement Learning Method for Continuous State Partially Observable Domains

Authors:
Yohannes Kassahun;Jose Gea;Jan Hendrik Metzen;Mark Edgington;Frank Kirchner
Affiliations:
Robotics Group, University of Bremen, Bremen, Germany D-28359;Robotics Group, University of Bremen, Bremen, Germany D-28359;German Research Center for Artificial Intelligence (DFKI), Bremen, Germany D-28359;Robotics Group, University of Bremen, Bremen, Germany D-28359;Robotics Group, University of Bremen, Bremen, Germany D-28359 and German Research Center for Artificial Intelligence (DFKI), Bremen, Germany D-28359
Venue:
KI '08 Proceedings of the 31st annual German conference on Advances in Artificial Intelligence
Year:
2008

Citing 8
Cited 0

Incremental evolution of complex general behavior

Adaptive Behavior - Special issue on environment structure and behavior
Efficient evolution of neural networks through complexification

Efficient evolution of neural networks through complexification
Completely Derandomized Self-Adaptation in Evolution Strategies

Evolutionary Computation
A common genetic encoding for both direct and indirect encodings of networks

Proceedings of the 9th annual conference on Genetic and evolutionary computation
Accelerating neuroevolutionary methods using a Kalman filter

Proceedings of the 10th annual conference on Genetic and evolutionary computation
Reinforcement learning: a survey

Journal of Artificial Intelligence Research
Neuroevolution with analog genetic encoding

PPSN'06 Proceedings of the 9th international conference on Parallel Problem Solving from Nature
Efficient non-linear control through neuroevolution

ECML'06 Proceedings of the 17th European conference on Machine Learning

Quantified Score

Hi-index	0.00

Visualization

Abstract

In this contribution we present an extension of a neuroevolutionary method called Evolutionary Acquisition of Neural Topologies (EANT) [11] that allows the evolution of solutions taking the form of a POMDP agent (Partially Observable Markov Decision Process) [8]. The solution we propose involves cascading a Kalman filter [10] (state estimator) and a feed-forward neural network. The extension (EANT+KALMAN) has been tested on the double pole balancing without velocity benchmark, achieving significantly better results than the to date published results of other algorithms.