On the Fairness of Disentangled Representations

Locatello, Francesco; Abbati, Gabriele; Rainforth, Tom; Bauer, Sebastian; Schölkopf, Bernhard; Bachem, Olivier

doi:10.48550/arxiv.1905.13662

Cited by 18 publications

(20 citation statements)

References 33 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Similar to separating content and style in natural images, disentanglement can potentially be used as an unsupervised approach for isolating protected attribution information such as ethnicity, which can be subsequently truncated from the latent code for downstream fairness tasks ?, 231 . Recent exhaustive studies on the evaluation of unsupervised VAE-based disentangled models have demonstrated that disentanglement scores correlate with fairness metrics, benchmarked on numerous fair classification tasks without protected attribute information 232 . In application to face identification, disentanglement-like methods have been proposed in clustering human faces without latent code information that contain dominant features such as as skin and hair color 233 .…”

Section: Fair Representation Learning Via Disentanglementmentioning

confidence: 99%

Algorithm Fairness in AI for Medicine and Healthcare

Chen¹,

Chen²,

Lipková³

et al. 2021

Preprint

View full text Add to dashboard Cite

In the current development and deployment of many artificial intelligence (AI) systems in healthcare, algorithm fairness is a challenging problem in delivering equitable care. Recent evaluation of AI models stratified across race sub-populations have revealed enormous inequalities in how patients are diagnosed, given treatments, and billed for healthcare costs. In this perspective article, we summarize the intersectional field of fairness in machine learning through the context of current issues in healthcare, outline how algorithmic biases (e.g. -image acquisition, genetic variation, intra-observer labeling variability) arise in current clinical workflows and their resulting healthcare disparities. Lastly, we also review emerging strategies for mitigating bias via decentralized learning, disentanglement, and model explainability.

show abstract

Section: Fair Representation Learning Via Disentanglementmentioning

confidence: 99%

Algorithm Fairness in AI for Medicine and Healthcare

Chen¹,

Chen²,

Lipková³

et al. 2021

Preprint

View full text Add to dashboard Cite

show abstract

“…Several debiasing methods have been proposed for existing models, the most simple being to remove subgroup indicators [10]. Others have used reinforcement learning methods to control the model disparity level [9,41], adversarial learning to generate a debiased input data [40,47,51], or target constructing a fair latent representation [1,24,26,50]. In this work we do not consider debiasing.…”

Section: Background and Related Workmentioning

confidence: 99%

CheXclusion: Fairness gaps in deep chest X-ray classifiers

Seyyed-Kalantari¹,

Liu²,

McDermott³

et al. 2020

Preprint

View full text Add to dashboard Cite

Machine learning systems have received much attention recently for their ability to achieve expert-level performance on clinical tasks, particularly in medical imaging. Here, we examine the extent to which state-of-the-art deep learning classifiers trained to yield diagnostic labels from X-ray images are biased with respect to protected attributes. We train convolution neural networks to predict 14 diagnostic labels in three prominent public chest X-ray datasets: MIMIC-CXR, Chest-Xray8, and CheXpert. We then evaluate the TPR disparity -the difference in true positive rates (TPR) andunderdiagnosis rate -the false positive rate of a non-diagnosisamong different protected attributes such as patient sex, age, race, and insurance type. We demonstrate that TPR disparities exist in the state-of-the-art classifiers in all datasets, for all clinical tasks, and all subgroups. We find that TPR disparities are most commonly not significantly correlated with a subgroup's proportional disease burden; further, we find that some subgroups and subsection of the population are chronically underdiagnosed. Such performance disparities have real consequences as models move from papers to products, and should be carefully audited prior to deployment. CCS CONCEPTS• Computing methodologies → Neural networks; Learning latent representations; • Applied computing → Imaging.

show abstract

“…This is demonstrably problematic, particularly because of the arbitrary nature of the functions being learned. It can result in bias against minority groups, and algorithmic decisions which are heavily influenced by culturally sensitive or legally protected characteristics such as race, age, gender, or sex (Hardt et al, 2016;Locatello et al, 2019;Cao and Daume III, 2019;Liu et al, 2019;Howard and Borenstein, 2018;Rose, 2010;Buolamwini and Gebru, 2018). In response to concern surrounding bias and opacity of ML driven decision Unfortunately, and as we will show, machine learning models and model explainability techniques cannot be used reliably to infer correlations, and the purview of the explainability should be restricted to be completely local to the model.…”

Section: Introductionmentioning

confidence: 99%

Trying to Outrun Causality with Machine Learning: Limitations of Model Explainability Techniques for Identifying Predictive Variables

Vowels¹

2022

Preprint

View full text Add to dashboard Cite

Machine Learning explainability techniques have been proposed as a means of 'explaining' or interrogating a model in order to understand why a particular decision or prediction has been made. Such an ability is especially important at a time when machine learning is being used to automate decision processes which concern sensitive factors and legal outcomes. Indeed, it is even a requirement according to EU law. Furthermore, researchers concerned with imposing overly restrictive functional form (e.g., as would be the case in a linear regression) may be motivated to use machine learning algorithms in conjunction with explainability techniques, as part of exploratory research, with the goal of identifying important variables which are associated with an outcome of interest. For example, epidemiologists might be interested in identifying 'risk factors' -i.e. factors which affect recovery from disease -by using random forests and assessing variable relevance using importance measures. However, and as we demonstrate, machine learning algorithms are not as flexible as they might seem, and are instead incredibly sensitive to the underling causal structure in the data. The consequences of this are that predictors which are, in fact, critical to a causal system and highly correlated with the outcome, may nonetheless be deemed by explainability techniques to be unrelated/unimportant/unpredictive of the outcome. Rather than this being a limitation of explainability techniques per se, we show that it is rather a consequence of the mathematical implications of regression, and the interaction of these implications with the associated conditional independencies of the underlying causal structure. We provide some alternative recommendations for researchers wanting to explore the data for important variables.

show abstract

On the Fairness of Disentangled Representations

Cited by 18 publications

References 33 publications

Algorithm Fairness in AI for Medicine and Healthcare

Algorithm Fairness in AI for Medicine and Healthcare

CheXclusion: Fairness gaps in deep chest X-ray classifiers

Trying to Outrun Causality with Machine Learning: Limitations of Model Explainability Techniques for Identifying Predictive Variables

Contact Info

Product

Resources

About