LSTM for time series forecasting with H20.ai

Question

I want to implement a time-series prediction model using LSTMs like the one mentioned here: https://machinelearningmastery.com/time-series-prediction-lstm-recurrent-neural-networks-python-keras/

The model works pretty well using a similar Keras code above, though I want to implement the same using H20.ai

This documentation here on H2O Deep Water says "The H2O Deep Water project supports CNNs and RNNs though third-party integrations of other deep learning libraries such as TensorFlow, Caffe and MXNet.". Though there are no demos for the same.

In this conversation here in Jan '16 on 'Recurrent Neural Networks in H2O for time series prediction?', they say it hasn't been implemented yet.

Any advice or links to similar implementations on H2O would be deeply appreciated

Note: I am looking for a data parallel architecture implementation rather than model parallelism

Answer 1

Active development on H2O's Deep Water project has stopped, see here: https://github.com/h2oai/deepwater

I don't think LSTM was included in the last release, and I would use Keras instead.

A list of Jupyter notebook demos is found here as well: https://github.com/h2oai/h2o-3/tree/master/examples/deeplearning/notebooks