Causality-based Neural Network Repair
Authors:
- Bing SunBSBing Sun
- Jun SunJSJun Sun
- Hong Long PhamHPHong Long Pham
- Jie ShiJSJie Shi
Preprint
Abstract
Neural networks have had discernible achievements in a wide range of applications. The wide-spread adoption also raises the concern of their dependability and reliability. Similar to traditional decision-making programs, neural networks can have defects that need to be repaired. The defects may cause unsafe behaviors, raise security concerns or unjust societal impacts. In this work, we address the problem of repairing a neural network for desirable properties such as fairness and the absence of backdoor. The goal is to construct a neural network that satisfies the property by (minimally) adjusting the given neural network's parameters (i.e., weights). Specifically, we propose CARE (\textbf{CA}usality-based \textbf{RE}pair), a causality-based neural network repair technique that 1) performs causality-based fault localization to identify the `guilty' neurons and 2) optimizes the parameters of the identified neurons to reduce the misbehavior. We have empirically evaluated CARE on various tasks such as backdoor removal, neural network repair for fairness and safety properties. Our experiment results show that CARE is able to repair all neural networks efficiently and effectively. For fairness repair tasks, CARE successfully improves fairness by $61.91\%$ on average. For backdoor removal tasks, CARE reduces the attack success rate from over $98\%$ to less than $1\%$. For safety property repair tasks, CARE reduces the property violation rate to less than $1\%$. Results also show that thanks to the causality-based fault localization, CARE's repair focuses on the misbehavior and preserves the accuracy of the neural networks.
Keywords
All Related Versions
This publication has 14 references indexed in Scilit:
- Towards Repairing Neural Networks Correctly2021
- Data augmentation in fault diagnosis based on the Wasserstein generative adversarial network with gradient penaltyNeurocomputing, 2020
- Causal Interpretations of Black-Box ModelsJournal of Business & Economic Statistics, 2019
- BadNets: Evaluating Backdooring Attacks on Deep Neural NetworksIEEE Access, 2019
- Formal Verification of Piece-Wise Linear Feed-Forward Neural Networks2017
- Reluplex: An Efficient SMT Solver for Verifying Deep Neural Networks2017
- Estimating Conditional Average Treatment EffectsJournal of Business & Economic Statistics, 2015
- Man vs. computer: Benchmarking machine learning algorithms for traffic sign recognitionNeural Networks, 2012
- What causes a system to satisfy a specification?ACM Transactions on Computational Logic, 2008
- Misunderstandings Between Experimentalists and Observationalists about Causal InferenceJournal of the Royal Statistical Society Series A: Statistics in Society, 2008