Mar 24, 2023 · It jointly smooths both input and weight loss landscapes in an adaptive, instance-specific, way to enhance robustness more for those samples ...

Mar 27, 2023 · Improved Adversarial Training Through Adaptive. Instance-wise Loss Smoothing. Lin Li and Michael Spratling. Abstract—Deep neural networks can ...

A new adversarial training method that jointly smooths both input and weight loss landscapes in an adaptive, instance-specific, way to enhance robustness ...

Mar 24, 2023 · It jointly smooths both input and weight loss landscapes in an adaptive, instance-specific, way to enhance robustness more for those samples ...

This repository contains the code of algorithm, Instance adaptive Smoothness Enhanced Adversarial Training (ISEAT), and pre-trained models from the paper " ...

We find that during training an overall reduction of adversarial loss is achieved by sacrificing a considerable proportion of training samples to be more ...

Adversarial training is by far the most successful strategy for improving robustness of neural networks to adversarial attacks.

Adversarial training has become one of the most effective methods for improving robustness of neu- ral networks. However, it often suffers from poor.

This paper proposed "surface smoothing adversarial training (SSAT)" that incorporates an adaptive label smoothing technique along with an adversarial training ...

Adversarial training has become one of the most effective methods for improving robustness of neural networks. However, it often suffers from poor ...