Robustness Verification

Robustness verification aims to formally guarantee the reliability of machine learning models, particularly neural networks, against adversarial attacks or noisy inputs. Current research focuses on developing efficient verification techniques for various architectures, including convolutional neural networks (CNNs), transformers, and recurrent neural networks (RNNs like LSTMs), often employing linear approximations, convex optimization, or set-based methods to analyze model behavior under perturbations. These advancements are crucial for deploying machine learning models in safety-critical applications, such as autonomous driving and healthcare, where reliable predictions are paramount. The field is actively pursuing tighter bounds and improved scalability to handle increasingly complex models and datasets.

Papers

June 14, 2024

Automated Design of Linear Bounding Functions for Sigmoidal Nonlinearities in Neural Networks
Matthias König, Xiyue Zhang, Holger H. Hoos, Marta Kwiatkowska, Jan N. van Rijn
Neural Network Native Robustness Activation Function Automated Design Robustness Verification Sigmoid Function Robustness Gap

June 2, 2024

Towards General Robustness Verification of MaxPool-based Convolutional Neural Networks via Tightening Linear Approximation
Yuan Xiao, Shiqing Ma, Juan Zhai, Chunrong Fang, Jinyuan Jia, Zhenyu Chen
Convolutional Neural Network Native Robustness Linear Function Approximation Robustness Verification

May 17, 2024

Boosting Few-Pixel Robustness Verification via Covering Verification Designs
Yuval Shapira, Naor Wiesel, Shahar Shabelman, Dana Drachsler-Cohen
Native Robustness Robustness Verification Local Robustness

March 20, 2024

Robustness Verifcation in Neural Networks
Adrian Wurm
Neural Network Formal Verification Robustness Verification Computational Challenge

March 6, 2024

Verified Training for Counterfactual Explanation Robustness under Data Shift
Anna P. Meyer, Yuhao Zhang, Aws Albarghouthi, Loris D'Antoni
Training Data Inherent Interpretability Counterfactual Explanation Robustness Verification Model Update Data Shift Efficiency Robustness

February 23, 2024

ProTIP: Probabilistic Robustness Verification on Text-to-Image Diffusion Models against Stochastic Perturbation
Yi Zhang, Yun Tang, Wenjie Ruan, Xiaowei Huang, Siddartha Khastgir, Paul Jennings, Xingyu Zhao
Diffusion Model Native Robustness Adversarial Example Text to Image Diffusion Model Text to Image Neural Architecture Robustness Verification Random Perturbation TIP Generation

February 22, 2024

Verifiable Boosted Tree Ensembles
Stefano Calzavara, Lorenzo Cazzaro, Claudio Lucchese, Giulio Ermanno Pibiri
Tree Ensemble Robustness Verification Verifiable Learning Fast Minimum Norm Attack

December 27, 2023

Robustness Verification for Knowledge-Based Logic of Risky Driving Scenes
Xia Wang, Anda Liang, Jonathan Sprinkle, Taylor T. Johnson
Data Driven Approach Knowledge Reasoning Driving Scene Robustness Verification Accident Datasets

December 15, 2023

Robustness Verification of Deep Reinforcement Learning Based Control Systems using Reward Martingales
Dapeng Zhi, Peixin Wang, Cheng Chen, Min Zhang
Deep Reinforcement Learning Robustness Verification Deep Reinforcement Learning Based Markovian Reward

November 16, 2023

Formal Verification of Long Short-Term Memory based Audio Classifiers: A Star based Approach
Neelanjana Pal, Taylor T Johnson
Long Short Term Memory Constructive Approach Audio Classification Formal Verification Signal Classification Star Model Robustness Verification Audio Classifier

August 11, 2023

TrajPAC: Towards Robustness Verification of Pedestrian Trajectory Prediction Models
Liang Zhang, Nathaniel Xu, Pengfei Yang, Gaojie Jin, Cheng-Chao Huang, Lijun Zhang
Native Robustness Trajectory Prediction Pedestrian Trajectory Prediction Robustness Verification State of the Art Trajectory

July 26, 2023

Robustness Verification of Deep Neural Networks using Star-Based Reachability Analysis with Variable-Length Time Series Input
Neelanjana Pal, Diego Manzanas Lopez, Taylor T Johnson
Deep Neural Network Native Robustness Reachability Analysis Deep Regression Robustness Verification Robustness Test

June 5, 2023

Evaluating robustness of support vector machines with the Lagrangian dual approach
Yuting Liu, Hong Gu, Pan Qin
Native Robustness Adversarial Robustness Support Vector Machine Lagrangian System Robustness Verification

May 23, 2023

Expressive Losses for Verified Robustness via Convex Combinations
Alessandro De Palma, Rudy Bunel, Krishnamurthy Dvijotham, M. Pawan Kumar, Robert Stanforth, Alessio Lomuscio
Adversarial Attack Adversarial Training Adversarial Robustness Expressive Power Robustness Verification Accuracy Robustness Convex Combination

March 9, 2023

March 3, 2023

Convex Bounds on the Softmax Function with Applications to Robustness Verification
Dennis Wei, Haoze Wu, Min Wu, Pin-Yu Chen, Clark Barrett, Eitan Farchi
Financial Application Softmax Function Lower Bound Robustness Verification Convex Formulation

February 2, 2023

Provably Bounding Neural Network Preimages
Suhas Kotha, Christopher Brix, Zico Kolter, Krishnamurthy Dvijotham, Huan Zhang
Adversarial Robustness Neural Network Verification Robustness Verification State of the Art Verifier

December 10, 2022

QVIP: An ILP-based Formal Verification Approach for Quantized Neural Networks
Yedi Zhang, Zhe Zhao, Fu Song, Min Zhang, Taolue Chen, Jun Sun
Deep Neural Network Quantization Operator Formal Verification Quantized Neural Network Robustness Verification Bit Quantization

November 13, 2022

Certifying Robustness of Convolutional Neural Networks with Tight Linear Approximation
Yuan Xiao, Tongtong Bai, Mingzheng Gu, Chunrong Fang, Zhenyu Chen
Convolutional Neural Network Native Robustness State of the Art Robustness Verification Tight Approximation Robustness Bound

Robustness Verification

Papers

Automated Design of Linear Bounding Functions for Sigmoidal Nonlinearities in Neural Networks

Towards General Robustness Verification of MaxPool-based Convolutional Neural Networks via Tightening Linear Approximation

Boosting Few-Pixel Robustness Verification via Covering Verification Designs

Robustness Verifcation in Neural Networks

Verified Training for Counterfactual Explanation Robustness under Data Shift

ProTIP: Probabilistic Robustness Verification on Text-to-Image Diffusion Models against Stochastic Perturbation

Verifiable Boosted Tree Ensembles

Robustness Verification for Knowledge-Based Logic of Risky Driving Scenes

Robustness Verification of Deep Reinforcement Learning Based Control Systems using Reward Martingales

Formal Verification of Long Short-Term Memory based Audio Classifiers: A Star based Approach

TrajPAC: Towards Robustness Verification of Pedestrian Trajectory Prediction Models

Robustness Verification of Deep Neural Networks using Star-Based Reachability Analysis with Variable-Length Time Series Input

Evaluating robustness of support vector machines with the Lagrangian dual approach

Expressive Losses for Verified Robustness via Convex Combinations

Efficient Certified Training and Robustness Verification of Neural ODEs

The Robustness Verification of Linear Sound Quantum Classifiers

Convex Bounds on the Softmax Function with Applications to Robustness Verification

Provably Bounding Neural Network Preimages

QVIP: An ILP-based Formal Verification Approach for Quantized Neural Networks

Certifying Robustness of Convolutional Neural Networks with Tight Linear Approximation