Enhancing adversarial attacks on text classification models through a modified word saliency-based approach.
