evaluate_class_mlp — Calculate the evaluation of a feature vector by a multilayer perceptron.
evaluate_class_mlp computes the result Result of evaluating the feature vector Features with the multilayer perceptron (MLP) MLPHandle. The formulas used for the evaluation are described with create_class_mlp. Before calling evaluate_class_mlp, the MLP must be trained with train_class_mlp.
If the MLP is used for regression (function approximation), i.e., if (OutputFunction = 'linear'), Result is the value of the function at the coordinate Features. For OutputFunction = 'logistic' and 'softmax', the values in Result can be interpreted as probabilities. Hence, for OutputFunction = 'logistic' the elements of Result represent the probabilities of the presence of the respective independent attributes. Typically, a threshold of 0.5 is used to decide whether the attribute is present or not. Depending on the application, other thresholds may be used as well. For OutputFunction = 'softmax' usually the position of the maximum value of Result is interpreted as the class of the feature vector, and the corresponding value as the probability of the class. In this case, classify_class_mlp should be used instead of evaluate_class_mlp because classify_class_mlp directly returns the class and corresponding probability.
Result of evaluating the feature vector with the MLP.
If the parameters are valid, the operator evaluate_class_mlp returns the value 2 (H_MSG_TRUE). If necessary, an exception is raised.
Christopher M. Bishop: “Neural Networks for Pattern Recognition”;
Oxford University Press, Oxford; 1995.
Andrew Webb: “Statistical Pattern Recognition”; Arnold, London; 1999.