PinnedNathan HubensinTowards Data ScienceDeep inside: AutoencodersAutoencoders (AE) are neural networks that aims to copy their inputs to their outputs. They work by compressing the input into a…Feb 25, 201822Feb 25, 201822
Nathan HubensinTowards Data ScienceFasterAI: a library to make smaller and faster neural networksFasterAI includes a collection of compression techniques for neural networks and is built on top of Fastai and PytorchNov 18, 20211Nov 18, 20211
Nathan HubensinTowards Data ScienceSpeed-up inference with Batch Normalization FoldingBatch Normalization is a technique which takes care of normalizing the input of each layer to make the training process faster and more…Apr 20, 20202Apr 20, 20202
Nathan HubensinTowards Data ScienceWhy default CNN are broken in Keras and how to fix themA deep dive into CNN initialization…Jul 6, 20198Jul 6, 20198
Nathan HubensinTowards Data ScienceTest Time Augmentation (TTA) and how to perform it with KerasData AugmentationFeb 11, 20194Feb 11, 20194
Nathan HubensinTowards Data ScienceBuild a simple Image Retrieval System with an AutoencoderImage retrieval is a very active and fast-advancing field of research area in the past decade. The most well-known systems being the Google…Aug 24, 20185Aug 24, 20185