Is it possible to save (and load) a trained LSTM network to generate C code out of it ?

Hi, I have trained a LSTM SeriesNetwork for time series data and want to generate C code out of it to implement it on a hardware. But the only functions I found are loadCompactModels for classification model, compact regression model, or nearest neighbor searcher and genFunction for FF nets and narxnet, but not for LSTMs. Does anybody know how this is possible? Thanks (I'm using Matlab 2018a)

Answer 1

[Surendar M]

No. I am using 2018b. Please search with 'Deep Learning Networks and Layers Supported for C++ Code Generation' in Matlab documentation and you will find the layers that are supported in deep learning for C/C++ code generation. LSTM is not listed in it.

-Surendar