Low-power implementation of Mitchell's approximate logarithmic multiplication for convolutional neural networks | Publicación