It's easy to use NCS Scalable Compute nodes to do Deep Learning with TensorFlow. Just spin up your secure ARM64 Scalable Compute instance, and execute the following commands to install TensorFlow:
curl -L -o - https://bootstrap.pypa.io/get-pip.py | python3 - --user
pip install --user --index-url https://pip.crystax.net/ tensorflow
Then you can upload your training data and training script (my_script.py) and begin training with the following command:
Coming Soon: Distributed Deep Learning at scale, using Neocortix Cloud Services.
And you can use NCS Scalable Compute nodes to do Federated Learning on mobile devices all over the world. Build machine learning models while protecting the privacy of local data, all with the TensorFlow Federated Learning framework.
Check out this friendly, comic-style introduction to Federated Learning from Google AI. It gradually builds up the key concepts, showing how "we can learn from everyone, without learning about anyone."
For greater depth, check out this paper on Federated Learning from the TensorFlow Federated team.
For further discussion, with some nice coding examples, check out this friendly video introduction.
Coming Soon: examples of using Federated Learning on Neocortix Cloud Services devices.