J.D.S. da Silva, H.F. de Campos Velho (2004): Artificial Neural Networks in Space Applications, Inaugural Symposium for the International Institute for Neuroscience of Natal (IINN), March 3-7, Natal, Brazil.

Abstract: Artificial neural networks (ANN) have emerged as an important tool for many applications in science and tecnology. The goal of this work is to present the use of ANNs in space science applications and space technology. The worked examples range from computer vision, satellite image classification, up to inverse problems solver (atmospheric temperature retrieval from satellite data). Many types of supervised and unsupervised ANNs have been employed in our studies, such as Multilayer Perceptron (with backpropagation learning process), Radial Basis Function, Cascade Correlation, LVQ (Learning Vector Quantization), Kohonen SOM (Self Organized Map), Hopfield, ART (Adaptive Resonance Theory), Hamming Net, BAM (Bidrectinal Associative Memories), Maxnet, FAM (Fuzzy Associative Memories).