(eng) Artificial neural networks may learn to solve arbitrary complex problems. But knowledge acquired is hard to exhibit. Thus neural networks appear as ``black boxes'', the decisions of which can't be explained. In this ...