In this paper, we propose a novel two-stage distortion-aware greedy-based sparse adversarial attack method “GreedyFool”. It can achieve both much better sparsity and invisibility than existing state-of-the-art sparse adversarial attack methods. It b…

We presented an implementation of Mask R-CNN with a GAN loss. The idea is to make the baseline mask generator as the the GAN generator, and then design a discriminator network to challenge these generations. This design obtains a loss that is learne…

The 2D human pose is widely used in tasks such as action or gesture recognition. However, for the detection of threats or dangerous objects such as firearms, only methods based on the visual appearance of the objects have been proposed, without taki…

In this paper, we propose a new information embedding attack that is verifiable and robust against post-processing of the classifier. In particular, our attack leverages CRC to verify the correctness of the recovered message. Moreover, our attack le…

This research has been financially supported by the National Institute of Standards and Technology (NIST) via the grant Pervasive, Accurate, and Reliable Location-based Services for Emergency Responders (Federal Grant: 70NANB17H185).

The uncertainty-equipped MaCC model that we present in this paper can be applied to many different scientific applications. We have shown in the case of large-scale numerical ICF simulations that it provides interpretable and meaningful uncertaintie…

In this work, we bound the change in output of spectral graph filters under a specific form of topological perturbation, i.e., edge rewiring, where the bound involves terms which have a structural interpretation. We then demonstrate a practical appl…