Apache Spark Deep Learning Cookbook

How it works...

A combination of a number of LSTM cells/units forms an LSTM network. The architecture of such a network is shown in the following diagram:

1. In the preceding diagram, the full LSTM cell is represented by "A". The cell takes the current input (x_i) of a sequence of inputs, and produces (h_i) which is nothing but the output of the current hidden state. This output is then sent to the next LSTM cell as its input.
2. An LSTM cell is slightly more complicated than an RNN cell. While the RNN cell has just one function/layer acting on a current input, the LSTM cell has three layers which are the three gates controlling the information flowing ...