Classification Using Multi-Layered Perceptrons

摘要: Abstract : There has been an increasing interest in the applicability of neural networks disparate domains. In this paper, we describe use multi-layered perceptrons, a type network topology, for financial classification problems, with promising results. Back-propagation, which is learning algorithm most often used however, inherently inefficient search procedure. We present improved procedures have much better convergence properties. Using several applications as examples, show efficacy using perceptrons algorithms. The modified algorithms performance, terms classification/prediction accuracies, than methods previously literature, such probit analysis and similarity-based techniques.