I am wondering if it is based on their own Google HyperTune / Vizier [1, 2] which is modified to better deal with uncertainties or it is an absolutely independent in-house development.
[1] https://cloud.google.com/ml-engine/docs/tensorflow/using-hyp...