- author = "Samy Bengio and Yoshua Bengio and Jocelyn Cloutier",
- title = "Use of genetic programming for the search of a new learning rule for neutral networks",
- booktitle = "Proceedings of the 1994 IEEE World Congress on Computational Intelligence",
- year = "1994",
- volume = "1",
- pages = "324--327",
- address = "Orlando, Florida, USA",
- month = "27-29 " # jun,
- publisher = "IEEE Press",
- keywords = "genetic algorithms, genetic programming, ANN",
- size = "4 pages",
-
URL = "
http://www.idiap.ch/~bengio/cv/publications/ps/bengio_1994_wcci.ps.gz",
-
URL = "
http://citeseer.ist.psu.edu/465154.html",
-
DOI = "
doi:10.1109/ICEC.1994.349932",
- abstract = "In previous work ([1, 2, 3]) we explained how to use standard optimization methods such as simulated annealing, gradient descent and genetic algorithms to optimize a parametric function which could be used as a learning rule for neural networks. To use these methods, we had to choose a fixed number of parameters and a rigid form for the learning rule. In this article, we propose to use genetic programming to find not only the values of rule parameters but also the optimal number of parameters and the form of the rule. Experiments on classification tasks suggest genetic programming finds better learning rules than other optimization methods. Furthermore, the best rule found with genetic programming outperformed the well-known backpropagation algorithm for a given set of tasks",
-
notes = "Uses GP to produce a learning rule for training a
neural network. Evolved rule like back-propergation but
better, differential is cubed. Says neural network is
fully connected,
",
