One word at a time: adversarial attacks on retrieval models

Nisarg Raval, Manisha Verma

Adversarial examples, generated by applying small perturbations to input features, are widely used to fool classifiers and measure their robustness to noisy inputs. However, little…