Approximation techniques for neuromimetic calculus

Vincent Vigneron (1, 2) , Claude Barret (2)

1 SAMOS - Statistique Appliquée et MOdélisation Stochastique
2 CEMIF - Centre d'Etude Mécanique d'Ile-de-France

Approximation Theory plays a central part in modern statistical methods, in particular in Neural Network modeling. These models are able to approximate a large amount of metric data structures in their entire range of definition or at least piecewise. We survey most of the known results for networks of neurone-like units. The connections to classical statistical ideas such as ordinary Least Squares are emphasized.

Mots clés: neural networks; ODE; LMS; optimization

Domaines: Autres [stat.ML]

Vincent Vigneron, Claude Barret. Approximation techniques for neuromimetic calculus. International Journal of Neural Systems, 1999, 9 (3), pp.227-234. ⟨hal-00201583⟩