Title :
Bayesian Training of Neural Networks Using Genetic Programming
Author :
Marwala, Tshilidzi
Author_Institution :
Witwatersrand Univ., Johannesburg
Abstract :
Bayesian neural networks trained using Markov chain Monte Carlo (MCMC) and genetic programming in binary space within Metropolis framework is proposed. It is tested and compared to classical MCMC method and is observed to give better results than classical approach.
Keywords :
Markov processes; Monte Carlo methods; belief networks; genetic algorithms; learning (artificial intelligence); multilayer perceptrons; Bayesian neural network training; Markov chain-Monte Carlo method; Metropolis framework; binary space; genetic programming; Bayesian methods; Distribution functions; Genetic programming; Helium; Inference algorithms; Monte Carlo methods; Multi-layer neural network; Neural networks; Probability distribution; Testing;
Conference_Titel :
Neural Networks, 2006. IJCNN '06. International Joint Conference on
Conference_Location :
Vancouver, BC
Print_ISBN :
0-7803-9490-9
DOI :
10.1109/IJCNN.2006.247374
