Rule Extraction From Binary Neural Networks With Convolutional Rules for Model Validation.

(These restarts did not seem to have an impact in the experiments we performed but may be significant in settings with larger step sizes or other datasets.) Therefore, one has to define a candidate space, a scoring function to evaluate a candidate solution, a neighborhood of a candidate solution, as well as a decision rule for selecting the next candidate out of a neighborhood.

also in 13,097 other papers