Parameter Aware Robustness

Parameter-aware robustness focuses on improving the resilience of machine learning models to various forms of noise and adversarial attacks by enhancing the robustness of their internal parameters. Current research emphasizes techniques like robust gradient projection, sharpness-aware minimization adaptations, and contrastive learning methods to achieve this, often applied within continual learning or differentially private settings. This research is crucial for building reliable and trustworthy AI systems, particularly in safety-critical applications where model performance must remain stable despite noisy or corrupted data. The resulting improvements in model stability and accuracy have significant implications for various fields, including autonomous driving and healthcare.

Papers

May 27, 2024

Improving Data-aware and Parameter-aware Robustness for Continual Learning
Hanxi Xiao, Fan Lyu
Native Robustness Continual LEArning Contrastive Loss Robust Continual Learning Parameter Aware Robustness

March 5, 2024

DPAdapter: Improving Differentially Private Deep Learning through Noise Tolerance Pre-training
Zihao Wang, Rui Zhu, Dongruo Zhou, Zhikun Zhang, John Mitchell, Haixu Tang, XiaoFeng Wang
Differentially Private Private Machine Learning Parameter Aware Robustness

October 16, 2023

SoTTA: Robust Test-Time Adaptation on Noisy Data Streams
Taesik Gong, Yewon Kim, Taeckyung Lee, Sorn Chottananurak, Sung-Ju Lee
Test Time Adaptation Noisy Data Continual Adaptation Time Adaptation Parameter Aware Robustness

February 17, 2023

Adversarial Contrastive Distillation with Adaptive Denoising
Yuzheng Wang, Zhaoyu Chen, Dingkang Yang, Yang Liu, Siao Liu, Wenqiang Zhang, Lizhe Qi
Contrastive Distillation General Adversarial Adversarial Robustness Distillation Parameter Aware Robustness

November 3, 2021

Pareto Adversarial Robustness: Balancing Spatial Robustness and Sensitivity-based Robustness
Ke Sun, Mingjie Li, Zhouchen Lin
Adversarial Robustness General Robustness Temporal Robustness Parameter Aware Robustness

Parameter Aware Robustness

Papers

Improving Data-aware and Parameter-aware Robustness for Continual Learning

DPAdapter: Improving Differentially Private Deep Learning through Noise Tolerance Pre-training

SoTTA: Robust Test-Time Adaptation on Noisy Data Streams

Adversarial Contrastive Distillation with Adaptive Denoising

Pareto Adversarial Robustness: Balancing Spatial Robustness and Sensitivity-based Robustness