domain-adversarial neural networks (dann)

Networks used for domain adaptation, where the model is trained to minimize the domain discrepancy while maximizing the domain classification accuracy.