Research lines

    1. New archtectures for Artificial neural network apllied to Forecast, classification and pattern recognition

    2. Auto-organizing artificial neural network with cybernetic principle, also called Constructive artificial neural networks. They are caracterized by using a hybrid methodology: artificial neural networks and statistic concepts. The main goal is the estimation of an optimal structure that evolves during the training process.

    3. New activation functioin, better normalization forms for better extrapolation, new objective functions

    4. Agorithms for artificial neural network training (backpropagation, swarm particle intelligence and others)

    5. Variable selection techniques (artificial neural networks combined with ant colonies optimization; harmonic search; artificial neural networks combined with temporal memory seach and others)