A Relaxed Optimization Approach for Adversarial Attacks against Neural Machine Translation Models

Open in new window