Feb 19, 2023 · In this paper, we propose a family of new losses, called stationary point (SP) loss, which has at least one stationary point on the correct classification side.

Feb 19, 2023 · Abstract—The inability to guarantee robustness is one of the major obstacles to the application of deep learning models in.

Feb 19, 2023 · A robust boundary should be kept in the middle of samples from different classes, thus maximizing the margins from the boundary to the samples.

It is proved that robust boundary can be guaranteed by SP loss without losing much accuracy, and it is demonstrated that robustness is improved under a ...

In this paper, we propose a family of new losses, called multi-stationary point (MS) loss, which introduce ad- ditional stationary points beyond the asymptotic ...

The inability to guarantee robustness is one of the major obstacles to the application of deep learning models in security-demanding domains.

The inability to guarantee robustness is one of the major obstacles to the application of deep learning models in security-demanding domains.

Stationary Point Losses for Robust Model ... CE loss sharpens the neural network at the decision boundary to achieve a lower loss, rather than pushing the ...

Distributionally robust optimization (DRO) is a widely-used approach to learn models that are robust against distribution shift. Compared with the standard.