Estimating and Improving Fairness with Adversarial Learning

Abstract

Fairness and accountability are two essential pillars for trustworthy Artificial Intelligence (AI) in healthcare. However, the existing AI model may be biased in its decision marking. To tackle this issue, we propose an adversarial multi-task training strategy to simultaneously mitigate and detect bias in the deep learning-based medical image analysis system. Specifically, we propose to add a discrimination module against bias and a critical module that predicts unfairness within the base classification model. We further impose an orthogonality regularization to force the two modules to be independent during training. Hence, we can keep these deep learning tasks distinct from one another, and avoid collapsing them into a singular point on the manifold. Through this adversarial training method, the data from the under-privileged group, which is vulnerable to bias because of attributes such as sex and skin tone, are transferred into a domain that is neutral relative to these attributes. Furthermore, the critical module can predict fairness scores for the data with unknown sensitive attributes. We evaluate our framework on a large-scale public-available skin lesion dataset under various fairness evaluation metrics. The experiments demonstrate the effectiveness of our proposed method for estimating and improving fairness in the deep learning-based medical image analysis system.

FUTURE-AI: Guiding Principles and Consensus Recommendations for Trustworthy Artificial Intelligence in Medical Imaging

“…8. Application of counter-measures: If biases are identified during any stage of development or testing of AI, mitigation measures should be investigated and evaluated, including (1) pre-processing approaches to improve the training dataset through re-sampling (under-or over-sampling), data augmentation (image synthesis using adversarial learning) or sample weighting to neutralise discriminatory effects; (2) in-processing approaches that modify the learning algorithm in order to remove discrimination during the model training process, such as by adding explicit constraints in the loss functions to minimise the performance difference between subgroups of individuals (e.g., learning bias-free representations via adversarial loss [96]); and (3) post-processing approaches to correct the outputs of the AI algorithm depending on the individual's group, such as by using the equalised odds post-processing technique [130]. All these techniques should be thoroughly evaluated to ensure their positive impact on fairness.…”

RadFusion: Benchmarking Performance and Fairness for Multimodal Pulmonary Embolism Detection from CT and EHR

“…In addition to overall model performance for different patient groups, we also assess the fairness of different models by reporting the equal opportunity difference (EOD) which measures the difference in TPR (i.e., Sensitivity) for the privileged and under-privileged groups following the evaluation protocol in [19,46,37]. We choose to use the TPR gap as our fairness metric based on the needs of the clinical diagnostic setting-high TPR disparity indicates that sick members from one demographic group would not be given correct diagnoses at the same rate as the general population, which can be dangerous for clinical deployment.…”

A Review of Generative Adversarial Networks in Cancer Imaging: New Applications, New Solutions

“…Aside from imbalance of labels, more insidious forms of imbalance such as that of race [82] or gender [83] of patients are easily omitted in studies. This leads to fairness problems in real world applications as underrepresenting such categories in the training set will hurt performance on these categories in the real world (population shift) [84]. Because of their potential to generate synthetic data, GANs are a promising solution to the aforementioned problems and have already been thoroughly explored in Computer Vision [85,86].…”

“…Towards the goal of a more diverse distribution of data with respect to gender and race, similar principles can be applied. For instance, Li et al [84] proposed an adversarial training scheme to improve fairness in classification of skin lesions for underrepresented groups (age, sex, skin tone) by learning a neutral representation using an adversarial bias discrimination loss. Fairness imposing GANs can also generate synthetic data with a preference for underrepresented groups, so that models may ingest a more bal-anced dataset, improving demographic parity without excluding data from the training pipeline.…”