Networks for images and sequences

Images and sequences have structure that a general network ignores. This topic slides small filters across images with convolutional layers, reads sequences step by step with recurrent networks, and builds attention and transformers from learned word vectors.