254,230 Hits in 3.4 sec

Generating Fair Universal Representations using Adversarial Models [article]

Peter Kairouz and Jiachun Liao and Chong Huang and Maunil Vyas and Monica Welfert and Lalitha Sankar
2022 arXiv   pre-print
We present a data-driven framework for learning fair universal representations (FUR) that guarantee statistical fairness for any learning task that may not be known a priori.  ...  For these datasets, we use two tradeoff measures: censoring vs. representation fidelity and fairness vs. utility for downstream tasks, to amply demonstrate that multiple sensitive features can be effectively  ...  CONCLUSION We have introduced an adversarial learning framework with verifiable guarantees for learning generative models that can create censored and fair universal representations for datasets with  ... 
arXiv:1910.00411v7 fatcat:mb3qtvly3ngp7csccbji3d6xny

Adversarial Stacked Auto-Encoders for Fair Representation Learning [article]

Patrik Joslin Kenfack, Adil Mehmood Khan, Rasheed Hussain, S.M. Ahsan Kazmi,
2021 arXiv   pre-print
Training machine learning models with the only accuracy as a final goal may promote prejudices and discriminatory behaviors embedded in the data.  ...  In this paper, we propose a new fair representation learning approach that leverages different levels of representation of data to tighten the fairness bounds of the learned representation.  ...  Learning fair representation is a model-agnostic approach to mitigate unfairness, i.e., the learned representation can be used for any downstream task and not only for neural network based models.  ... 
arXiv:2107.12826v1 fatcat:dpajjw43dfbrfgdtyaox2luzsq

Learning Fair Representations for Recommendation: A Graph-based Perspective [article]

Le Wu, Lei Chen, Pengyang Shao, Richang Hong, Xiting Wang, Meng Wang
2021 arXiv   pre-print
Most of these approaches assumed independence of instances, and designed sophisticated models to eliminate the sensitive information to facilitate fairness.  ...  Specifically, given the original embeddings from any recommendation models, we learn a composition of filters that transform each user's and each item's original embeddings into a filtered embedding space  ...  ACKNOWLEDGEMENTS This work was supported in part by grants from the National Natural Science Foundation of China (Grant No. 61972125, U19A2079, U1936219, 61932009, 91846201), and CAAI-Huawei MindSpore  ... 
arXiv:2102.09140v3 fatcat:eul7bvqyrzbjjb5z7yg2p5d2vu

Learning Fair and Transferable Representations [article]

Luca Oneto, Michele Donini, Andreas Maurer, Massimiliano Pontil
2020 arXiv   pre-print
We derive learning bounds establishing that the learned representation transfers well to novel tasks both in terms of prediction performance and fairness metrics.  ...  One way to reach this goal is by modifying the data representation in order to meet certain fairness constraints. In this work we measure fairness according to demographic parity.  ...  By this we mean that when the representation is used to learn novel tasks, it is guaranteed to learn a model that has both a small error and meets the fairness requirement.  ... 
arXiv:1906.10673v3 fatcat:4tv32aypgfgspbdrx7i5r2lx2m

Learning Adversarially Fair and Transferable Representations [article]

David Madras, Elliot Creager, Toniann Pitassi, Richard Zemel
2018 arXiv   pre-print
utility, an essential goal of fair representation learning.  ...  Furthermore, we present the first in-depth experimental demonstration of fair transfer learning and demonstrate empirically that our learned representations admit fair predictions on new tasks while maintaining  ...  This work was supported by the Canadian Institute for Advanced Research (CIFAR) and the Natural Sciences and Engineering Research Council of Canada (NSERC).  ... 
arXiv:1802.06309v3 fatcat:fcp32kts3jg63nqkjdsudquvxm

Modeling Techniques for Machine Learning Fairness: A Survey [article]

Mingyang Wan, Daochen Zha, Ninghao Liu, Na Zou
2022 arXiv   pre-print
mitigate fairness issues in outputs and representations.  ...  focuses on refining latent representation learning.  ...  The implicit approaches often target deep learning models, where learning representations is crucial.  ... 
arXiv:2111.03015v2 fatcat:didcuo2yabbcrb2fuhveqgng3y

Learning to Ignore: Fair and Task Independent Representations [article]

Linda H. Boedi, Helmut Grabner
2021 arXiv   pre-print
We apply it to learn fair models and interpret the influence of the sensitive attribute.  ...  Training fair machine learning models, aiming for their interpretability and solving the problem of domain shift has gained a lot of interest in the last years.  ...  In this paper we propose a yet simple approach for learning fair representation.  ... 
arXiv:2101.04047v2 fatcat:s7otfesvsjhelfz672heupny3m

Fair Representation Learning using Interpolation Enabled Disentanglement [article]

Akshita Jha, Bhanukiran Vinzamuri, Chandan K. Reddy
2021 arXiv   pre-print
In this paper, we propose a novel method to address two key issues: (a) Can we simultaneously learn fair disentangled representations while ensuring the utility of the learned representation for downstream  ...  To address the former, we propose the method FRIED, Fair Representation learning using Interpolation Enabled Disentanglement.  ...  Learning Fair and Disentangled Representations The inputs, X and p are fed into the encoder f , which generates the latent representations for the input instances (X 1 , p) and (X 2 , p) by passing it  ... 
arXiv:2108.00295v2 fatcat:mqfgkxzbkfb2vj4naeqvl6no4a

Fair Interpretable Learning via Correction Vectors [article]

Mattia Cerrato and Marius Köppel and Alexander Segner and Stefan Kramer
2022 arXiv   pre-print
Neural network architectures have been extensively employed in the fair representation learning setting, where the objective is to learn a new representation for a given vector which is independent of  ...  We show experimentally that a fair representation learning problem constrained in such a way does not impact performance.  ...  Commonly, the learning of fair representations is achieved by learning a new feature space Z starting from the input space X.  ... 
arXiv:2201.06343v1 fatcat:s7kbxinwsva2zfjoqy5eymwnvq

Fairness in Deep Learning: A Computational Perspective [article]

Mengnan Du, Fan Yang, Na Zou, Xia Hu
2020 arXiv   pre-print
We also discuss fairness mitigation approaches categorized according to three stages of deep learning life-cycle, aiming to push forward the area of fairness in deep learning and build genuinely fair and  ...  However, deep learning models might exhibit algorithmic discrimination behaviors with respect to protected groups, potentially posing negative impacts on individuals and society.  ...  Among different machine learning models, the fairness problem of deep learning models has attracted attention from academia and industry recently.  ... 
arXiv:1908.08843v2 fatcat:kaaevm64fbctpjfdycv5uz3dhi

FairRec: Fairness-aware News Recommendation with Decomposed Adversarial Learning [article]

Chuhan Wu, Fangzhao Wu, Xiting Wang, Yongfeng Huang, Xing Xie
2021 arXiv   pre-print
Existing news recommendation models are usually learned from users' news click behaviors.  ...  ., genders) have similar patterns and news recommendation models can easily capture these patterns.  ...  The news and user models in our approach are based on the neural news and user models in the NRMS [21] method. The news model learns news representations from news titles.  ... 
arXiv:2006.16742v2 fatcat:ajuqjjwowvgahn6pda5zrfv24y

FairNN- Conjoint Learning of Fair Representations for Fair Decisions [article]

Tongxin Hu, Vasileios Iosifidis, Wentong Liao, Hang Zhang, Michael YingYang, Eirini Ntoutsi, Bodo Rosenhahn
2020 arXiv   pre-print
In this paper, we propose FairNN a neural network that performs joint feature representation and classification for fairness-aware learning.  ...  loss and (c) allows for solving a classification task in a fair manner by minimizing the classification error and respecting the equalized odds-based fairness regularized.  ...  In [26] an approach for learning individually fair representations is proposed using an end-to-end model with autoencoders.  ... 
arXiv:2004.02173v2 fatcat:byo2lzu6ynbwrmsaje74ytezvq

On the Fairness of Disentangled Representations [article]

Francesco Locatello, Gabriele Abbati, Tom Rainforth, Stefan Bauer, Bernhard Schölkopf, Olivier Bachem
2019 arXiv   pre-print
Recently there has been a significant interest in learning disentangled representations, as they promise increased interpretability, generalization to unseen scenarios and faster learning on downstream  ...  Analyzing the representations of more than 12600 trained state-of-the-art disentangled models, we observe that several disentanglement scores are consistently correlated with increased fairness, suggesting  ...  Acknowledgements The authors thank Sylvain Gelly and Niki Kilbertus for helpful discussions and comments.  ... 
arXiv:1905.13662v2 fatcat:rgvgcl62kzctblzho65pcrvyzy

README: REpresentation learning by fairness-Aware Disentangling MEthod [article]

Sungho Park, Dohyung Kim, Sunhee Hwang, Hyeran Byun
2020 arXiv   pre-print
In this paper, we design Fairness-aware Disentangling Variational AutoEncoder (FD-VAE) for fair representation learning.  ...  Fair representation learning aims to encode invariant representation with respect to the protected attribute, such as gender or age.  ...  Acknowledgments and Disclosure of Funding TBD.  ... 
arXiv:2007.03775v1 fatcat:nbliab6mpfcy7an5xifqfmtfcu

Learning Fair Representations via an Adversarial Framework [article]

Rui Feng, Yang Yang, Yuehan Lyu, Chenhao Tan, Yizhou Sun, Chunping Wang
2019 arXiv   pre-print
., race and gender), and tackle this problem by learning latent representations of individuals that are statistically indistinguishable between protected groups while sufficiently preserving other information  ...  Our framework provides a theoretical guarantee with respect to statistical parity and individual fairness.  ...  Fair representation learning. Most relevant to our work are studies on fair representation learning. Zemel et al.  ... 
arXiv:1904.13341v1 fatcat:pou3ms3enzhndnj5fqzdieegi4
« Previous Showing results 1 — 15 out of 254,230 results