DeMem can be seamlessly integrated into various adver- sarial training techniques

We propose DeMem, a method that selectively restricts high-risk samples to enhance privacy while minimizing performance degradation

1 Pith paper cite this work. Polarity classification is still indexing.

1 Pith paper citing it

browse 1 citing papers

representative citing papers

cs.LG · 2024-12-08 · conditional · novelty 5.0

Adding a loss-variance penalty to adversarial training reduces membership inference leakage on CIFAR-10/100 while keeping accuracy mostly intact.

Showing 1 of 1 citing paper.

DeMem: Privacy-Enhanced Robust Adversarial Learning via De-Memorization cs.LG · 2024-12-08 · conditional · none · ref 4
Adding a loss-variance penalty to adversarial training reduces membership inference leakage on CIFAR-10/100 while keeping accuracy mostly intact.