Warning: This paper contains text examples that are offensive or harmful in nature.

Specifically, we randomly sample an example and adopt a first-order procedure to approximate the Hessian/vector product, which makes computing more efficient by interpolating two neighboring gradients.

In other words, the word order may undergo changes as a result and leads to a low text similarity.

Recently, text-to-image models have been thriving. Despite their powerful generative capacity, our research has uncovered a lack of robustness in this generation process.

Neural Information Processing Systems http://nips.cc/

Existing attacks almost perform on low-level instances, such as pixels and super-pixels, and rarely exploit semantic clues.

Like most NAS methods in AutoML, the discrete selection in the perturbation block is more interpretable and robust (e.g., L1-Norm for feature selection and single path in NAS) than the mixture