Solving Time of Least Square Systems in Sigma-Pi Unit Networks

Pierre Courrieu (1)

1 LPC - Laboratoire de psychologie cognitive

The solving of least square systems is a useful operation in neurocomputational modeling of learning, pattern matching, and pattern recognition. In these last two cases, the solution must be obtained on-line, thus the time required to solve a system in a plausible neural architecture is critical. This paper presents a recurrent network of Sigma-Pi neurons, whose solving time increases at most like the logarithm of the system size, and of its condition number, which provides plausible computation times for biological systems.

Mots clés: Sigma-Pi Neurons; Recurrent Neural Network; Least Square Systems; On-line Pattern Matching; RBFN Learning

Domaines: Réseau de neurones [cs.NE]

Pierre Courrieu. Solving Time of Least Square Systems in Sigma-Pi Unit Networks. Neural Information Processing - Letters and Reviews, 2004, 4 (3), pp.39-45. ⟨hal-00276480⟩