An input-output map in which the patterns are divided into classes is considered for the perceptron. The statistical mechanical analysis with a finite number of classes turns out to give the same results as the case of only one class of patterns; the limit of capacity and the relevant order parameters are calculated in a mean field approach. The analysis is then extended to the Derrida Gardner canonical ensemble in which the perceptron can be studied beyond the limit of capacity. We complete the analysis with numerical simulations with the perceptron learning rule. The revelance of those results to the possible emergence of spontaneous categorization is finally discussed Nous considerons une application entree-sortie pour un perceptron dans laquelle les formes sont divisees en classe. L'analyse par la mecanique statistique dans le cas d'un nombre fini de classe donne les memes resultats que pour une seule classe; nous calculons la limite de capacite et les parametres d'ordre pertinents en champ moyen. Nous generalisons ensuite l'analyse a l'ensemble canonique de Derrida-Gardner dans lequel le perceptron peut etre etudie au-dela de sa limite de capacite. Nous completons l'analyse en etudiant numeriquement la regle d'apprentissage du perceptron. Nous discutons finalement la relevance de ces resutlats a l'emergence possible d'une categorisation spontanee
[1]
A. A. Mullin,et al.
Principles of neurodynamics
,
1962
.
[2]
J J Hopfield,et al.
Neural networks and physical systems with emergent collective computational abilities.
,
1982,
Proceedings of the National Academy of Sciences of the United States of America.
[3]
Néstor Parga,et al.
The ultrametric organization of memories in a neural network
,
1986
.
[4]
S Dehaene,et al.
Spin glass model of learning by selection.
,
1986,
Proceedings of the National Academy of Sciences of the United States of America.
[5]
James L. McClelland,et al.
Parallel distributed processing: explorations in the microstructure of cognition, vol. 1: foundations
,
1986
.
[6]
P. Carnevali,et al.
Exhaustive Thermodynamical Analysis of Boolean Learning Networks
,
1987
.
[7]
E. Gardner,et al.
Maximum Storage Capacity in Neural Networks
,
1987
.
[8]
Lev B. Ioffe,et al.
The Augmented Models of Associative Memory Asymmetric Interaction and Hierarchy of Patterns - Int. J. Mod. Phys. B1, 51 (1987)
,
1987
.
[9]
Stefano Patarnello,et al.
Learning Networks of Neurons with Boolean Logic
,
1987
.
[10]
E. Gardner,et al.
Optimal storage properties of neural network models
,
1988
.
[11]
E. Gardner.
The space of interactions in neural network models
,
1988
.