Balanced Adversarial Training: Balancing Tradeoffs between Fickleness and Obstinacy in NLP Models

Hannah Chen, Yangfeng Ji, David Evans