Data Augmentation via Subgroup Mixup for Improving Fairness

Data Augmentation via Subgroup Mixup for Improving Fairness - arXiv

Data augmentation for group fairness allows us to add new samples of underrepresented groups to balance subpopulations. Furthermore, our method ...

Data Augmentation via Subgroup Mixup for Improving Fairness - ar5iv

1 Introduction · (1). We perform data augmentation via the efficient and effective pairwise mixup to improve the fairness and accuracy of model predictions; · ( ...

Data Augmentation via Subgroup Mixup for Improving Fairness

... It is intended to be an evaluation dataset for detecting demographic bias in ASR systems. Navarro et al. [22] propose a data augmentation ...

data augmentation via subgroup mixup for improving fairness

Madeline Navarro, Camille Little, Genevera Allen, Santiago Segarra, Rice University, United States of America ...

ICLR 2021, Fair Mixup: Fairness via Interpolation - GitHub

To improve the generalizability of fair classifiers, we propose fair mixup, a new data augmentation strategy for imposing the fairness constraint. In ...

GBMix: Enhancing Fairness by Group-Balanced Mixup - IEEE Xplore

Mixup is a powerful data augmentation strategy that has been shown to improve the generalization and adversarial robustness of machine ...

FAIR MIXUP: FAIRNESS VIA INTERPOLATION - OpenReview

To improve the generalizability of fair classifiers, we propose fair mixup, a new data augmentation strategy for imposing the fairness constraint. In partic-.

(PDF) GBMix: Enhancing Fairness by Group-Balanced Mixup

PDF | Mixup is a powerful data augmentation strategy that has been shown to improve the generalization and adversarial robustness of machine learning.

May 23, 2024 – The paper presents a data augmentation method designed to improve intersectional fairness in classification tasks by leveraging the ...

Global Mixup: Eliminating Ambiguity with Clustering

Data augmentation with Mixup ... a new paradigm for data augmentation, through split sample ... Auto- matic data augmentation has improved significant perfor-.

NSF Award Search: Award # 2210837 - Minipatch Learning for ...

... data poses major statistical and computational challenges. ... through evaluation using the Foundation's ... Augmentation via Subgroup Mixup for Improving Fairness ...

ProxiMix: Enhancing Fairness with Proximity Samples in Subgroups

Segarra, Data augmentation via subgroup mixup for improving fairness, arXiv preprint arXiv:2309.07110 (2023). [11] C.-Y. Chuang, Y. Mroueh, Fair mixup ...

‪Camille Little‬ - ‪Google Scholar‬

Data Augmentation via Subgroup Mixup for Improving Fairness. M Navarro, C Little, GI Allen, S Segarra. ICASSP 2024-2024 IEEE International Conference on ...

Mixup-based Data Augmentation for Differentially Private Learning

... improving the generalization performance and the adversarial robustness. similar · inspect. -75.09. Data Augmentation via Subgroup Mixup for Improving Fairness.

RC-Mixup: A Data Augmentation Strategy against Noisy Data for ...

At the same time, robust training has been heavily studied where the goal is to train accurate models against noisy data through multiple rounds ...

GBMix: Enhancing Fairness by Group-Balanced Mixup - IEEE Xplore

ABSTRACT Mixup is a powerful data augmentation strategy that has been shown to improve the generalization and adversarial robustness of ...

[PDF] Fair Mixup: Fairness via Interpolation | Semantic Scholar

Fair mixup, a new data augmentation strategy for imposing the fairness ... Data Augmentation via Subgroup Mixup for Improving Fairness · Madeline ...

Anchor Data Augmentation - arxiv-sanity

In this work, we propose data augmentation via pairwise mixup across subgroups to improve group fairness. Many real-world applications of machine learning ...

Sep 13, 2023 – This research proposes a data augmentation technique called subgroup mixup to improve fairness in machine learning systems.

PatchMix: patch-level mixup for data augmentation in convolutional ...

... Data augmentation via subgroup mixup for improving fairness. In ICASSP 2024-2024 IEEE international conference on acoustics, speech and signal processing ...