White-box Fairness Testing through Adversarial Sampling

@article{Zhang2020WhiteboxFT,
  title={White-box Fairness Testing through Adversarial Sampling},
  author={Peixin Zhang and Jingyi Wang and Jun Sun and Guoliang Dong and Xinyu Wang and Xingen Wang and Jin Song Dong and Ting Dai},
  journal={2020 IEEE/ACM 42nd International Conference on Software Engineering (ICSE)},
  year={2020},
  pages={949-960}
}
  • Peixin Zhang, Jingyi Wang, Ting Dai
  • Published 27 June 2020
  • Computer Science
  • 2020 IEEE/ACM 42nd International Conference on Software Engineering (ICSE)
Although deep neural networks (DNNs) have demonstrated astonishing performance in many applications, there are still concerns on their dependability. One desirable property of DNN for applications with societal impact is fairness (i.e., non-discrimination). In this work, we propose a scalable approach for searching individual discriminatory instances of DNN. Compared with state-of-the-art methods, our approach only employs lightweight procedures like gradient computation and clustering, which… 
Automatic Fairness Testing of Neural Classifiers through Adversarial Sampling
TLDR
This work proposes a scalable and effective approach for systematically searching for discriminative samples while extending fairness testing to address a challenging domain, i.e., text classification.
NeuronFair: Interpretable White-Box Fairness Testing through Biased Neuron Identification
TLDR
NeuronFair is a new DNN fairness testing framework that differs from previous work in several key aspects: interpretable it quantitatively interprets DNNs’ fairness violations for the biased decision; effective it uses the interpretation results to guide the generation of more diverse instances in less time; generic it can handle both structured and unstructured data.
SOCRATES: Towards a Unified Platform for Neural Network Verification
TLDR
SOCRATES is a unified framework for developing verification techniques for neural networks which supports a standardized format for a variety of neural network models, an assertion language for property specification as well as two novel algorithms for verifying or falsifying neural networks models.
Fairness Testing of Machine Learning Models Using Deep Reinforcement Learning
  • Wentao Xie, Peng Wu
  • Computer Science
    2020 IEEE 19th International Conference on Trust, Security and Privacy in Computing and Communications (TrustCom)
  • 2020
TLDR
This paper presents a systematic approach of testing the fairness of a machine learning model, with individual discriminatory inputs generated automatically in an adaptive manner based on the state-of-the-art deep reinforcement learning techniques.
FairNeuron: Improving Deep Neural Network Fairness with Adversary Games on Selective Neurons
TLDR
FairNeuron, a DNN model automatic repairing tool, is proposed to mitigate fairness concerns and balance the accuracy-fairness trade-off without introducing another model, and evaluation on three datasets shows that FairNeuron can effectively improve all models’ fairness while maintaining a stable utility.
Fairness Testing of Deep Image Classification with Adequacy Metrics
TLDR
DeepFAIT consists of several important components enabling effective fairness testing of deep image classification applications: a neuron selection strategy to identify the fairnessrelated neurons, a set of multi-granularity adequacy metrics to evaluate the model’s fairness, and a test selection algorithm for fixing the fairness issues efficiently.
Probabilistic Verification of Neural Networks Against Group Fairness
TLDR
This work proposes an approach to formally verify neural networks against fairness, with a focus on independence-based fairness such as group fairness, and demonstrates that with the analysis results, the neural weights can be optimized to improve fairness.
Fairness Matters - A Data-Driven Framework Towards Fair and High Performing Facial Recognition Systems
TLDR
A novel approach to mitigate unfairness and enhance performance of state-of-the-art face recognition algorithms, using distribution-aware dataset curation and augmentation, and a 4-fold increase in fairness towards ethnicity when compared to related work is presented.
Fairness-aware Configuration of Machine Learning Libraries
TLDR
This paper designs three search-based software testing algorithms to uncover the precision-fairness frontier of the hyperparameter space of machine learning algorithms, and implements the proposed approaches in the tool Parfait-ML, which shows its effectiveness and utility over five mature ML algorithms as used in six social-critical applications.
Explanation-Guided Fairness Testing through Genetic Algorithm
TLDR
Experiments show that ExpGA presents higher efficiency and effectiveness than four state-of-the-art approaches, and only requires prediction probabilities of the tested model, resulting in a better generalization capability to various models.
...
1
2
3
4
...

References

SHOWING 1-10 OF 28 REFERENCES
Adversarial Sample Detection for Deep Neural Network through Model Mutation Testing
TLDR
This work proposes a measure of 'sensitivity' and shows empirically that normal samples and adversarial samples have distinguishable sensitivity, and integrates statistical hypothesis testing and model mutation testing to check whether an input sample is likely to be normal or adversarial at runtime by measuring its sensitivity.
Explaining and Harnessing Adversarial Examples
TLDR
It is argued that the primary cause of neural networks' vulnerability to adversarial perturbation is their linear nature, supported by new quantitative results while giving the first explanation of the most intriguing fact about them: their generalization across architectures and training sets.
Data Decisions and Theoretical Implications when Adversarially Learning Fair Representations
TLDR
An adversarial training procedure is used to remove information about the sensitive attribute from the latent representation learned by a neural network, and the data distribution empirically drives the adversary's notion of fairness.
Automated Directed Fairness Testing
TLDR
It is shown that Aequitas effectively generates inputs to uncover fairness violation in all the subject classifiers and systematically improves the fairness of respective models using the generated test inputs.
The Limitations of Deep Learning in Adversarial Settings
TLDR
This work formalizes the space of adversaries against deep neural networks (DNNs) and introduces a novel class of algorithms to craft adversarial samples based on a precise understanding of the mapping between inputs and outputs of DNNs.
DeepXplore: Automated Whitebox Testing of Deep Learning Systems
TLDR
DeepXplore efficiently finds thousands of incorrect corner case behaviors in state-of-the-art DL models with thousands of neurons trained on five popular datasets including ImageNet and Udacity self-driving challenge data.
FairTest: Discovering Unwarranted Associations in Data-Driven Applications
TLDR
The unwarranted associations (UA) framework is introduced, a principled methodology for the discovery of unfair, discriminatory, or offensive user treatment in data-driven applications and instantiate the UA framework in FairTest, the first comprehensive tool that helps developers check data- driven applications for unfair user treatment.
Fairness through awareness
TLDR
A framework for fair classification comprising a (hypothetical) task-specific metric for determining the degree to which individuals are similar with respect to the classification task at hand and an algorithm for maximizing utility subject to the fairness constraint, that similar individuals are treated similarly is presented.
Adversarial examples in the physical world
TLDR
It is found that a large fraction of adversarial examples are classified incorrectly even when perceived through the camera, which shows that even in physical world scenarios, machine learning systems are vulnerable to adversarialExamples.
Satisfying Real-world Goals with Dataset Constraints
TLDR
This paper proposes handling multiple goals on multiple datasets by training with dataset constraints, using the ramp penalty to accurately quantify costs, and presents an efficient algorithm to approximately optimize the resulting non-convex constrained optimization problem.
...
1
2
3
...