Recurrent neural network on convolution neural network

A recurrent neural network can be combined with convolutional neural network features to produce new sentences. This enables end-to-end training of the models. The following is the architecture of such a model:

LSTM model (Source: https://arxiv.org/pdf/1411.4555.pdf)

There are several layers of LSTM used to produce the desired results. A few of the results produced by this model are shown in the following screenshot:

Source: https://arxiv.org/pdf/1411.4555.pdf

These results are better ...

Get TensorFlow Deep Learning Projects now with the O’Reilly learning platform.

O’Reilly members experience books, live events, courses curated by job role, and more from O’Reilly and nearly 200 top publishers.