Margin Discrepancy-based Adversarial Training for Multi-Domain Text Classification

Open in new window