A Hierarchical Classification of First-Order Recurrent Neural Networks - Inserm - Institut national de la santé et de la recherche médicale Access content directly
Conference Papers Year : 2011

A Hierarchical Classification of First-Order Recurrent Neural Networks


We provide a refined hierarchical classification of first-order recurrent neural networks made up of McCulloch and Pitts cells. The classification is achieved by first proving the equivalence between the expressive powers of such neural networks and Muller automata, and then translating the Wadge classification theory from the automata-theoretic to the neural network context. The obtained hierarchical classification of neural networks consists of a decidable pre-well ordering of width 2 and height !!, and a decidability procedure of this hierarchy is provided. Notably, this classification is shown to be intimately related to the attractive properties of the networks, and hence provides a new refined measurement of the computational power of these networks in terms of their attractive behaviours.
Fichier principal
Vignette du fichier
Cabessa_2010_A-hierarchical_Auteur.pdf (1.7 Mo) Télécharger le fichier
Origin Files produced by the author(s)

Dates and versions

inserm-00624083 , version 1 (15-09-2011)



Jérémie Cabessa, Alessandro E. Villa. A Hierarchical Classification of First-Order Recurrent Neural Networks. 4th International Conference on Language and Automata Theory and Applications, May 2011, Trier, Germany. pp.142-153, ⟨10.1007/978-3-642-13089-2_12⟩. ⟨inserm-00624083⟩


115 View
475 Download



Gmail Mastodon Facebook X LinkedIn More