J.A. Carrasco-Ochoa et al. (Eds.): MCPR 2012, LNCS 7329, pp. 23–34, 2012. © Springer-Verlag Berlin Heidelberg 2012 Automatic Design of Artificial Neural Networks and Associative Memories for Pattern Classification and Pattern Restoration Humberto Sossa 1 , Beatriz A. Garro 1 , Juan Villegas 2 , Carlos Avilés 2 , and Gustavo Olague 3 1 CIC-IPN, Juan de Dios Batiz S/N, Col. Nva. Industrial Vallejo, México, D. F. Mexico 2 UAM-Azcapotzalco, Av. San Pablo Xalpa 180. Azcapotzalco, México, D. F. Mexico 3 CICESE, Carretera Ensenada-Tijuana 3918 Zona Playitas, Ensenada, B. C., Mexico hsossa@cic.ipn.mx, {beatriz.auroragl,jvillegas,gustavo.olague}@gmail.com, caviles@correo.azc.uam.mx Abstract. In this note we present our most recent advances in the automatic de- sign of artificial neural networks (ANNs) and associative memories (AMs) for pattern classification and pattern recall. Particle Swarm Optimization (PSO), Differential Evolution (DE), and Artificial Bee Colony (ABC) algorithms are used for ANNs; Genetic Programming is adopted for AMs. The derived ANNs and AMs are tested with several examples of well-known databases. As we will show, results are very promising. Keywords: Artificial neural networks, Associative memories, Evolutionary programming. 1 Introduction Pattern recognition (PR) is a main topic in machine vision. If we want a machine to efficiently interact with its environment, it is necessary that the above mentioned problem is correctly solved. Many approaches to face this problem have been reported in literature. One of the most popular is the artificial neural network based approach. It consists on combining many small processors (programs) in such a way that a set of patterns under study is correctly classified or restored. An artificial neural network (ANN) can be seen as a set of highly interconnected processors. The processors can be electronic devices or computer programs. From now on, these processors will be called nodes or units. These units can be the nodes of a graph. The edges of this graph determine the interconnections among the nodes. These represent the synaptic connections between the nodes, and are supposed to be similar to the synaptic connections between biological neurons of a brain. Associative memories, in the other hand, are special cases of ANNs. They have sev- eral interesting properties that make them preferable than ANNs, for some problems.