The paper offers a critical analysis of the procedure observed in many applications of neural networks. Given a problem to be solved, a favorite NN-architecture is chosen and its p...
We demonstrate the advantages of using Bayesian multi layer perceptron (MLP) neural networks for image analysis. The Bayesian approach provides consistent way to do inference by c...
We have proposed a complete system for text detection and localization in gray scale scene images. A boosting framework integrating feature and weak classifier selection based on...
Independent Component Analysis (ICA) is a statistical computation method that transforms a random vector in another one whose components are independent. Because the marginal distr...
Jorge Igual, Jehad I. Ababneh, Raul Llinares, Juli...
Creating ensembles of random but "realistic" topologies for complex systems is crucial for many tasks such as benchmark generation and algorithm analysis. In general, exp...