Neural Network Robustness

Neural network robustness research focuses on improving the resilience of deep learning models to various forms of input perturbations, including adversarial attacks and naturally occurring corruptions, aiming to enhance reliability and safety in real-world applications. Current research explores diverse techniques such as adversarial training, data augmentation (including label augmentation), and architectural modifications across various model architectures (e.g., convolutional neural networks, vision transformers). This field is crucial for deploying reliable AI systems in safety-critical domains (e.g., autonomous driving, medical diagnosis) and for advancing our understanding of how to build more trustworthy and generalizable AI models.

Papers

March 30, 2023

ImageNet-E: Benchmarking Neural Network Robustness via Attribute Editing
Xiaodan Li, Yuefeng Chen, Yao Zhu, Shuhui Wang, Rong Zhang, Hui Xue
Native Robustness Supervised ImageNet Neural Network Robustness ImageNet E Attribute Editing

March 17, 2023

Robust Mode Connectivity-Oriented Adversarial Defense: Enhancing Neural Network Robustness Against Diversified $\ell_p$ Attacks
Ren Wang, Yuxuan Li, Sijia Liu
Adversarial Attack Adversarial Robustness Adversarial DEfense Neural Network Robustness Norm Attack \Ell_p$ Robustness

November 28, 2022

Establishment of Neural Networks Robust to Label Noise
Pengwei Yang, Chongyangzi Teng, Jack George Mangos
Noisy Label Label Noise Neural Network Robustness Noise Transition Matrix Label Noise Robustness

November 17, 2022

Assessing Neural Network Robustness via Adversarial Pivotal Tuning
Peter Ebert Christensen, Vésteinn Snæbjarnarson, Andrea Dittadi, Serge Belongie, Sagie Benaim
Image Classifier Neural Network Robustness Adversarial Fine Tuning Semantic Image Editing

July 5, 2022

PRoA: A Probabilistic Robustness Assessment against Functional Perturbations
Tianle Zhang, Wenjie Ruan, Jonathan E. Fieldsend
Native Robustness Small Perturbation Neural Network Robustness Probabilistic Robustness

June 21, 2022

Why Robust Natural Language Understanding is a Challenge
Marco Casadio, Ekaterina Komendantskaya, Verena Rieser, Matthew L. Daggitt, Daniel Kienitz, Luca Arnaboldi, Wen Kokke
Deep Learning Native Robustness Natural Language Challenge Task Neural Network Robustness Robust Natural Language Understanding

June 13, 2022

Towards Alternative Techniques for Improving Adversarial Robustness: Analysis of Adversarial Training at a Spectrum of Perturbations
Kaustubh Sridhar, Souradeep Dutta, Ramneet Kaur, James Weimer, Oleg Sokolsky, Insup Lee
Native Robustness General Analysis Adversarial Training Adversarial Robustness Adversarial Perturbation Additive Perturbation Neural Network Robustness Sensitivity Aware

April 24, 2022

Improving Deep Learning Model Robustness Against Adversarial Attack by Increasing the Network Capacity
Marco Marchetti, Edmond S. L. Ho
Deep Learning Adversarial Attack Native Robustness Deep Learning Algorithm Neural Network Robustness Model Resilience

January 31, 2022

LinSyn: Synthesizing Tight Linear Bounds for Arbitrary Neural Network Activation Functions
Brandon Paulsen, Chao Wang
Native Robustness Activation Function Robustness Certification Neural Network Robustness

January 5, 2022

Corrupting Data to Remove Deceptive Perturbation: Using Preprocessing Method to Improve System Robustness
Hieu Le, Hans Walker, Dung Tran, Peter Chin
Deep Neural Network Generative Adversarial Network Neural Network Robustness Preprocessing Method Data Corruption Perturbation Artifact

December 1, 2021

A Systematic Review of Robustness in Deep Learning for Computer Vision: Mind the gap?
Nathan Drenkow, Numair Sani, Ilya Shpitser, Mathias Unberath
Deep Learning Native Robustness Computer Vision Adversarial Robustness Systematic Review Neural Network Robustness Robustness Issue

Neural Network Robustness

Papers

ImageNet-E: Benchmarking Neural Network Robustness via Attribute Editing

Robust Mode Connectivity-Oriented Adversarial Defense: Enhancing Neural Network Robustness Against Diversified $\ell_p$ Attacks

Establishment of Neural Networks Robust to Label Noise

Assessing Neural Network Robustness via Adversarial Pivotal Tuning

PRoA: A Probabilistic Robustness Assessment against Functional Perturbations

Why Robust Natural Language Understanding is a Challenge

Towards Alternative Techniques for Improving Adversarial Robustness: Analysis of Adversarial Training at a Spectrum of Perturbations

Improving Deep Learning Model Robustness Against Adversarial Attack by Increasing the Network Capacity

LinSyn: Synthesizing Tight Linear Bounds for Arbitrary Neural Network Activation Functions

Corrupting Data to Remove Deceptive Perturbation: Using Preprocessing Method to Improve System Robustness

A Systematic Review of Robustness in Deep Learning for Computer Vision: Mind the gap?