It’s well known that the difference in executable size between a compiled binary and one hand-written in optimized assembler ...

One of the greatest limitations of deep neural networks is the difficulty of interpreting what they learn from the data. Deep distilling addresses this problem by extracting human-comprehensible and ...