I had been looking at some OCR problems and came across this presentation. I implemented it.

In the presentation, there is the LSTM-Stack (diagram and algorithm, slide 32):

enter image description here

Here is a visualization of my LSTM stack in Keras:

enter image description here

I actually made my own attention model and I concatenated the attention part of the input for the 1st LSTM layer. In the hope of it adding some kind of context.

I am training the neural network for a couple of days so far.

Is this a good implementation? Please, give me any suggestions.



You must log in to answer this question.

Browse other questions tagged .