F3: Fair and Federated Face Attribute Classification with Heterogeneous Data

Kanaparthy, Samhita; Manisha, Padala,; Damle, Sankarshan; Sarvadevabhatla, Ravi Kiran; Gujar, Sujit

doi:10.48550/arxiv.2109.02351

Cited by 2 publications

(3 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Abay et al [1] analyzes multiple causes of unfairness in federated learning and potential solutions. Other papers that consider notions related to egalitarian fairness include [20,21,15,9].…”

Section: Fairness Definitions and Ethical Considerationsmentioning

confidence: 99%

Models of fairness in federated learning

Donahue¹,

Kleinberg²

2021

Preprint

View full text Add to dashboard Cite

In many real-world situations, data is distributed across multiple locations and can't be combined for training. Federated learning is a novel distributed learning approach that allows multiple federating agents to jointly learn a model. While this approach might reduce the error each agent experiences, it also raises questions of fairness: to what extent can the error experienced by one agent be significantly lower than the error experienced by another agent? In this work, we consider two notions of fairness that each may be appropriate in different circumstances: egalitarian fairness (which aims to bound how dissimilar error rates can be) and proportional fairness (which aims to reward players for contributing more data). For egalitarian fairness, we obtain a tight multiplicative bound on how widely error rates can diverge between agents federating together. For proportional fairness, we show that sub-proportional error (relative to the number of data points contributed) is guaranteed for any individually rational federating coalition.

show abstract

Section: Fairness Definitions and Ethical Considerationsmentioning

confidence: 99%

Models of fairness in federated learning

Donahue¹,

Kleinberg²

2021

Preprint

View full text Add to dashboard Cite

show abstract

“…each client represents a bank with data about female as well as male customers) Mohri et al [2019], Abay et al [2020], Du et al [2021], Cui et al [2021] and/or by sacrificing some privacy for fairness. Recently proposed methods (Kanaparthy et al [2021], Papadaki et al [2021], Yue et al [2021]) send unprotected model parameters and/or other unprotected information such as fairness metrics to a central aggregator, thereby leaking information about clients' data Boenisch et al [2021]. FairFed Ezzeldin et al [2021] employs techniques to protect the individual updates but the aggregations at the central aggregator are not protected and can reveal information about the clients' data when under attack.…”

Section: Introductionmentioning

confidence: 99%

“…For example, during FL training, assuming that an adversary A has the model from the previous round and the gradient updates from the current round, A can infer a private training exampleKairouz et al [2021]. Current works do not take into account such information leaks in FLPapadaki et al [2021],Yue et al [2021],Kanaparthy et al [2021]. A can also analyze the aggregated outputs to infer knowledge about a particular client.…”

mentioning

confidence: 99%

PrivFairFL: Privacy-Preserving Group Fairness in Federated Learning

Pentyala¹,

Neophytou²,

Nascimento³

et al. 2022

Preprint

View full text Add to dashboard Cite

Group fairness ensures that the outcome of machine learning (ML) based decision making systems are not biased towards a certain group of people defined by a sensitive attribute such as gender or ethnicity. Achieving group fairness in Federated Learning (FL) is challenging because mitigating bias inherently requires using the sensitive attribute values of all clients, while FL is aimed precisely at protecting privacy by not giving access to the clients' data. As we show in this paper, this conflict between fairness and privacy in FL can be resolved by combining FL with Secure Multiparty Computation (MPC) and Differential Privacy (DP). In doing so, we propose a method for training group-fair ML models in cross-device FL under complete and formal privacy guarantees, without requiring the clients to disclose their sensitive attribute values. Empirical evaluations on real world datasets demonstrate the effectiveness of our solution to train fair and accurate ML models in federated cross-device setups with privacy guarantees to the users.

show abstract

F3: Fair and Federated Face Attribute Classification with Heterogeneous Data

Cited by 2 publications

References 13 publications

Models of fairness in federated learning

Models of fairness in federated learning

PrivFairFL: Privacy-Preserving Group Fairness in Federated Learning

Contact Info

Product

Resources

About