no code implementations • 5 Jan 2017 • Hongjun Lu, Rudy Setiono, Huan Liu
One of the major reasons cited is that knowledge generated by neural networks is not explicitly represented in the form of rules suitable for verification or interpretation by humans.