Recurrent Neural Networks (RNNs) are neural networks that are designed for predicting sequence data. Images are not traditionally seen as sequence data, but can be modeled as such. Today we’re going to be testing out how well three different RNN architectures, Simple RNNs, LSTMs, and GRUs, do on image classification via the MNIST digits dataset.Continue reading “The Best RNN for Image Classification: RNN, LSTM, or GRU?”