work Disentangling Symbols and Movements Factor-VAE on NAR Dataset for few-shot visual analogical reasoning Reading Noisy Captions Embedded in Images Encoder-Decoder network for extracting text from images using attention-based LSTM fun