Complexity search for compressed neural networks

Faustino Gomez, Jan Koutník, Jürgen Schmidhuber

Research output: Chapter in Book/Report/Conference proceedingConference contribution


In this paper, we introduce a method, called Compressed Network Complexity Search (CNCS), for automatically determining the complexity of compressed networks (neural networks encoded indirectly by Fourier-type coefficients) that favors parsimonious solutions. CNCS maintains a probability distribution over complexity classes that it uses to select which class to optimize. Class probabilities are adapted based on their expected fitness, starting with a prior biased toward the simplest networks. Experiments on a challenging non-linear version of the helicopter hovering task, show that the method consistently finds simple solutions. Copyright is held by the author/owner(s).
Original languageEnglish (US)
Title of host publicationGECCO'12 - Proceedings of the 14th International Conference on Genetic and Evolutionary Computation Companion
PublisherAssociation for Computing Machinery
Number of pages2
ISBN (Print)9781450311786
StatePublished - Jan 1 2012
Externally publishedYes

Bibliographical note

Generated from Scopus record by KAUST IRTS on 2022-09-14


Dive into the research topics of 'Complexity search for compressed neural networks'. Together they form a unique fingerprint.

Cite this