Label-Free Supervision of Neural Networks with Physics and Domain Knowledge
The recipient of this year’s Outstanding Paper Award utilizes prior domain knowledge to constrain output space to a specific learning structure rather than a simple mapping from input to output