Bv

B. van Zwienen

1 records found

In the literature, neural network compression can significantly reduce the number of floating-point operations (FLOPs) of a neural network with limited accuracy loss. At the same time, it is common to manually design smaller networks instead of using modern compression techniques ...