Simple and near-optimal algorithms for hidden stratification and multi-group learning

Tosh, Christopher; Hsu, Daniel

doi:10.48550/arxiv.2112.12181

Cited by 1 publication

(2 citation statements)

References 23 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Concurrently and independently of our paper, Tosh and Hsu [Tosh and Hsu, 2021] study algorithms and sample complexity for multi-group agnostic learnability and give an algorithm ("Prepend") that is equivalent to our Algorithm 1 ("ListUpdate"). Their focus is on sample complexity of batch optimization, howeverin contrast to our focus on the discovery of groups on which our model is performing poorly online (e.g.…”

Section: Related Workmentioning

confidence: 99%

“…They also are not concerned with the details of the optimization that needs to be solved to produce an update -we give practical algorithms based on reductions to cost sensitive classification and empirical evaluation. Tosh and Hsu [Tosh and Hsu, 2021] also contains additional results, including algorithms producing more complex hypotheses but with improved sample complexity (again in the setting in which the groups are fixed up front).…”

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

An Algorithmic Framework for Bias Bounties

Globus-Harris¹,

Kearns²,

Roth³

2022

Preprint

View full text Add to dashboard Cite

Notions of fair machine learning that seek to control various kinds of error across protected groups generally are cast as constrained optimization problems over a fixed model class. For all such problems, tradeoffs arise: asking for various kinds of technical fairness requires compromising on overall error, and adding more protected groups increases error rates across all groups. Our goal is to "break though" such accuracy-fairness tradeoffs, also known as Pareto frontiers.We develop a simple algorithmic framework that allows us to deploy models and then revise them dynamically when groups are discovered on which the error rate is suboptimal. Protected groups do not need to be specified ahead of time: At any point, if it is discovered that there is some group on which our current model is performing substantially worse than optimally, then there is a simple update operation that improves the error on that group without increasing either overall error, or the error on any previously identified group. We do not restrict the complexity of the groups that can be identified, and they can intersect in arbitrary ways. The key insight that allows us to break through the tradeoff barrier is to dynamically expand the model class as new high error groups are identified. The result is provably fast convergence to a model that cannot be distinguished from the Bayes optimal predictorat least by the party tasked with finding high error groups.We explore two instantiations of this framework: as a "bias bug bounty" design in which external auditors are invited (and monetarily incentivized) to discover groups on which our current model's error is suboptimal, and as an algorithmic paradigm in which the discovery of groups on which the error is suboptimal is posed as an optimization problem. In the bias bounty case, when we say that a model cannot be distinguished from Bayes optimal, we mean by any participant in the bounty program. We provide both theoretical analysis and experimental validation.

show abstract

Section: Related Workmentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%