This paper presents initial results of Generalized Compressed Network Search (GCNS), a method for automatically identifying the important frequencies for neural networks encoded as a set of Fourier-type coeficients (i.e. \compressed" networks ). GCNS achieves better compression than our previous approach, and promises better generalization capabilities. Results for a high-dimensional Octopus arm control problem show that a high fitness 3680-weight network can be encoded using less than 10 coeficients, using the frequencies identified by GCNS. Copyright is held by the author/owner(s).
|Title of host publication
|GECCO'12 - Proceedings of the 14th International Conference on Genetic and Evolutionary Computation Companion
|Association for Computing Machinery
|Number of pages
|Published - Jan 1 2012