DNN Robustness

Deep neural network (DNN) robustness research focuses on improving the resilience of DNNs to various forms of perturbation, including adversarial attacks and data distribution shifts. Current efforts explore diverse strategies, such as incorporating noise into model architectures (e.g., NoisyGNNs), developing novel training algorithms (e.g., negative feedback training), and employing architectural search methods to design inherently robust models (e.g., for Graph Neural Networks). These advancements are crucial for ensuring the reliability and safety of DNNs in high-stakes applications like autonomous driving and healthcare, where unpredictable inputs or environmental changes could have severe consequences.

Papers

April 17, 2024

Exploring DNN Robustness Against Adversarial Attacks Using Approximate Multipliers
Mohammad Javad Askarizadeh, Ebrahim Farahmand, Jorge Castro-Godinez, Ali Mahani, Laura Cabrera-Quiros, Carlos Salazar-Garcia
Deep Neural Network Adversarial Attack DNN Robustness Approximate Multiplier

March 25, 2024

DeepKnowledge: Generalisation-Driven Deep Learning Testing
Sondess Missaoui, Simos Gerasimou, Nikolaos Matragkas
Adversarial Attack Depth Information Deep Learning Testing DNN Robustness DNN Testing

February 21, 2024

A Simple and Yet Fairly Effective Defense for Graph Neural Networks
Sofiane Ennadir, Yassine Abbahaddou, Johannes F. Lutzeyer, Michalis Vazirgiannis, Henrik Boström
Graph Neural Network Defense Method Adversarial Defense DNN Robustness

January 15, 2024

Robustness Against Adversarial Attacks via Learning Confined Adversarial Polytopes
Shayan Mohajer Hamidi, Linfeng Ye
Adversarial Attack Native Robustness Adversarial Robustness DNN Robustness Adversarial Polytope

July 29, 2023

Improving Realistic Worst-Case Performance of NVCiM DNN Accelerators through Training with Right-Censored Gaussian Noise
Zheyu Yan, Yifan Qin, Wujie Wen, Xiaobo Sharon Hu, Yiyu Shi
Training Data Gaussian Noise Worst Case DNN Accelerator DNN Robustness Compute in Memory

May 23, 2023

Negative Feedback Training: A Novel Concept to Improve Robustness of NVCIM DNN Accelerators
Yifan Qin, Zheyu Yan, Wujie Wen, Xiaobo Sharon Hu, Yiyu Shi
Native Robustness DNN Accelerator DNN Inference Non Volatile Memory DNN Robustness Novel Concept Device Variation Negative Training

April 20, 2023

Using Z3 for Formal Modeling and Verification of FNN Global Robustness
Yihao Zhang, Zeming Wei, Xiyue Zhang, Meng Sun
Adversarial Robustness Verification Task Formal Model Global Deep Global Robustness DNN Robustness Adversarial Region

April 9, 2023

Adversarially Robust Neural Architecture Search for Graph Neural Networks
Beini Xie, Heng Chang, Ziwei Zhang, Xin Wang, Daixin Wang, Zhiqiang Zhang, Rex Ying, Wenwu Zhu
Graph Neural Network GNN Architecture DNN Robustness Robust Graph Neural Network Graph Adversarial Attack Robust Neural Architecture Search

February 9, 2023

Et Tu Certifications: Robustness Certificates Yield Better Adversarial Examples
Andrew C. Cullen, Shijie Liu, Paul Montague, Sarah M. Erfani, Benjamin I. P. Rubinstein
Adversarial Example Attack Performance DNN Robustness Robustness Certificate

November 18, 2022

A Tale of Two Cities: Data and Configuration Variances in Robust Deep Learning
Guanqin Zhang, Jiankun Sun, Feng Xu, H. M. N. Dilum Bandara, Shiping Chen, Yulei Sui, Tim Menzies
Deep Learning Deep Neural Network Native Robustness Raw Data Urban Environment DNN Model Robust Deep Cautionary TALE DNN Robustness Parameter Configuration

September 29, 2022

Generalizability of Adversarial Robustness Under Distribution Shifts
Kumail Alhamoud, Hasan Abed Al Kader Hammoud, Motasem Alfarra, Bernard Ghanem
Native Robustness Adversarial Robustness Distribution Shift Stronger Generalizability Certified Robustness Adversarial Augmentation DNN Robustness

July 9, 2022

Training Robust Deep Models for Time-Series Domain: Novel Algorithms and Theoretical Analysis
Taha Belkhouja, Yan Yan, Janardhan Rao Doppa
Theoretical Analysis Robust Training Robust Deep Novel Algorithm DNN Robustness Time Series Alignment

June 20, 2022

Understanding Robust Learning through the Lens of Representation Similarities
Christian Cianfarani, Arjun Nitin Bhagoji, Vikash Sehwag, Ben Y. Zhao, Prateek Mittal, Haitao Zheng
Representation Learning Camera Lens Robust Learning Robust Representation Robust Training Similar Representation DNN Robustness Robust Network

May 20, 2022

The developmental trajectory of object recognition robustness: children are like small adults but unlike big deep neural networks
Lukas S. Huber, Robert Geirhos, Felix A. Wichmann
Neural Network Object Recognition Nine Year Old Child DNN Robustness Developmental Trajectory

March 21, 2022

Benchmarking Test-Time Unsupervised Deep Neural Network Adaptation on Edge Devices
Kshitij Bhardwaj, James Diffenderfer, Bhavya Kailkhura, Maya Gokhale
Edge Device Benchmark Platform Batch Normalization Device Deep Resource Constrained Edge DNN Robustness

January 30, 2022

GARNET: Reduced-Rank Topology Learning for Robust and Scalable Graph Neural Networks
Chenhui Deng, Xiuyu Li, Zhuo Feng, Zhiru Zhang
Graph Neural Network Adversarial Attack Adversarial Robustness State of the Art Robust Version Topological Learning DNN Robustness Adversarial Feature Meta )Stable GARNET System

DNN Robustness

Papers

Exploring DNN Robustness Against Adversarial Attacks Using Approximate Multipliers

DeepKnowledge: Generalisation-Driven Deep Learning Testing

A Simple and Yet Fairly Effective Defense for Graph Neural Networks

Robustness Against Adversarial Attacks via Learning Confined Adversarial Polytopes

Improving Realistic Worst-Case Performance of NVCiM DNN Accelerators through Training with Right-Censored Gaussian Noise

Negative Feedback Training: A Novel Concept to Improve Robustness of NVCIM DNN Accelerators

Using Z3 for Formal Modeling and Verification of FNN Global Robustness

Adversarially Robust Neural Architecture Search for Graph Neural Networks

Et Tu Certifications: Robustness Certificates Yield Better Adversarial Examples

A Tale of Two Cities: Data and Configuration Variances in Robust Deep Learning

Generalizability of Adversarial Robustness Under Distribution Shifts

Training Robust Deep Models for Time-Series Domain: Novel Algorithms and Theoretical Analysis

Understanding Robust Learning through the Lens of Representation Similarities

The developmental trajectory of object recognition robustness: children are like small adults but unlike big deep neural networks

Benchmarking Test-Time Unsupervised Deep Neural Network Adaptation on Edge Devices

GARNET: Reduced-Rank Topology Learning for Robust and Scalable Graph Neural Networks