This example shows how you would provision a pre-trained network.  TripleBlind
supports Keras .h5 models, PyTorch .pth models, but the preferred format is
the Open Neural Network Exchange .onnx format.  Once a trained network is
positioned on the Access Point, it can be used just like any model trained
within the TripleBlind system.  Agreements can be created allowing other
organizations to use the trained model.

The remote inference sample code show how another organization could infer
against this trained model with their own data.  SMPC inference is illustrated.

NOTE: Keras support has been deprecated due to changes in recent versions.  It
has been commented out from this example, but could be revived if there is
strong need for it.
