Learning to map between ferns with differentiable binary embedding networks

Max Blendowski, Mattias P. Heinrich

Current deep learning methods are based on the repeated, expensive application of convolutions with parameter-intensive weight matrices. In this work, we present a novel concept th…