Statistical Models and Artificial Neural Networks

摘要: A class of artificial neural networks (ANN) are interpreted as complex multivariate statistical models for the approximation an unknown expectation function a random variable y given explanatory x. Thus, weights these ANN can be viewed parameters, which estimated by methods. Important network compared with equivalent models. The closeness is evaluated distance functions. Quasi Maximum Likelihood and nonlinear least squares methods used estimation methods, depending on chosen function. Estimation algorithms (Newton-Raphson, weight decay, adaptive stepsize, generalized delta rule, etc.) discussed some significance tests described.