Extracting Rules from Neural Networks with Partial Interpretations

Cosimo Persia, Ana Ozaki

We investigate the problem of extracting rules, expressed in Horn logic, from neural network models. Our work is based on the exact learning model, in which a learner interacts wit…