Heinrich Braun

Learn More
A new learning algorithm for multilayer feedforward networks, RPROP, is proposed. To overcome the inherent disadvantages of pure gradient-descent, RPROP performs a local adaptation of the weight-updates according to the behaviour of the errorfunction. In substantial difference to other adaptive techniques, the effect of the RPROP adaptation process is not(More)
ENZO-M combines two successful search techniques using two diierent timescales: learning (gradient descent) for netuning of each oospring and evolution for coarse optimization steps of the network topology. Therefore, our evolutionary algorithm is a metaheuristic based on the best available local heuristic. Through training each oospring by fast gradient(More)
For many practical problem domains the use of neural networks has led to very satisfactory results. Nevertheless the choice of an appropriate, problem specific network architecture still remains a very poorly understood task. Given an actual problem, one can choose a few different architectures, train the chosen architectures a few times and finally select(More)
As an introduction to heuristics based on the evolution as a paradigm, we give an overview over the most well known types: genetic algorithms, genetic programming, evolution strategies and evolutionary programming. Since the neural network paradigm seems to be a most promising approach to implement the intelligence of an agent (i.e. its control strategy),(More)
In this paper we study the combination of two powerful approaches, evolutionary topology optimization (ENZO) and Tempoal Diierence Learning (TD()) which is up to our knowledge the rst time. Temporal Diierence Learning was proven to be a well suited technique for learning strategies for solving reinforcement problems based on neural network models , whereas(More)