Filters








2,883 Hits in 3.1 sec

Abstracting Fairness: Oracles, Metrics, and Interpretability

Cynthia Dwork, Christina Ilvento, Guy N. Rothblum, Pragya Sur, Aaron Roth
2020 Symposium on Foundations of Responsible Computing  
The oracle takes as input a (context, classifier) pair satisfying an arbitrary fairness definition, and accepts or rejects the pair according to whether the classifier satisfies the underlying fairness  ...  Since every "truly fair" classifier induces a coarse metric, in which those receiving the same decision are at distance zero from one another and those receiving different decisions are at distance one  ...  The data analysts 8:8 Abstracting Fairness: Oracles, Metrics, and Interpretability have no knowledge of M whatsoever, but hope to learn about fair classifiers by interacting with the oracle.  ... 
doi:10.4230/lipics.forc.2020.8 dblp:conf/forc/DworkIRS20 fatcat:dc2ahgm5wbhu7j3nmynnx2vq3y

Abstracting Fairness: Oracles, Metrics, and Interpretability [article]

Cynthia Dwork, Christina Ilvento, Guy N. Rothblum, Pragya Sur
2020 arXiv   pre-print
The oracle takes as input a (context, classifier) pair satisfying an arbitrary fairness definition, and accepts or rejects the pair according to whether the classifier satisfies the underlying fairness  ...  Since every "truly fair" classifier induces a coarse metric, in which those receiving the same decision are at distance zero from one another and those receiving different decisions are at distance one  ...  We abstract the problem by instantiating "true fairness" (and the expert who knows what this is) via an oracle that holds a collection T ⊆ {0, 1} |X | of vectors specifying the classification outcome for  ... 
arXiv:2004.01840v1 fatcat:6rcrsu7favcq7mlsuhull3base

Quantum Fair Machine Learning [article]

Elija Perrier
2021 arXiv   pre-print
, metrics and remediation strategies when quantum algorithms are subject to fairness constraints.  ...  We extend canonical Lipschitz-conditioned individual fairness criteria to the quantum setting using quantum metrics.  ...  ACKNOWLEDGMENTS This work is part of the author's research into quantum ethics at the ANU Humanising Machine Intelligence programme and Centre for Quantum Software and Information at UTS, Sydney.  ... 
arXiv:2102.00753v2 fatcat:wpvmnxfyx5bebbqw3z5ckdqt6e

Failure detectors encapsulate fairness

Scott M. Pike, Srikanth Sastry, Jennifer L. Welch
2012 Distributed computing  
Instead, we argue that failure detectors encapsulate the fairness in computation and communication.  ...  We argue that partially synchronous systems are perhaps better specified with fairness constraints (rather than real-time constraints) on computation and communication.  ...  eventually accurate oracles and perpetually accurate oracles.  ... 
doi:10.1007/s00446-012-0164-x fatcat:66sjj7obv5h4logv42kqjoc6h4

Failure Detectors Encapsulate Fairness [chapter]

Scott M. Pike, Srikanth Sastry, Jennifer L. Welch
2010 Lecture Notes in Computer Science  
Instead, we argue that failure detectors encapsulate the fairness in computation and communication.  ...  We argue that partially synchronous systems are perhaps better specified with fairness constraints (rather than real-time constraints) on computation and communication.  ...  eventually accurate oracles and perpetually accurate oracles.  ... 
doi:10.1007/978-3-642-17653-1_15 fatcat:5psifzlq3fdujimlv7tkqqzipy

Software Fairness: An Analysis and Survey [article]

Ezekiel Soremekun, Mike Papadakis, Maxime Cordy, Yves Le Traon
2022 arXiv   pre-print
This includes specifying, designing, and validating fairness properties.  ...  Our findings include the following: (1) Fairness concerns (such as fairness specification and requirements engineering) are under-studied; (2) Fairness measures such as conditional, sequential, and intersectional  ...  , Analysis detecting confusion and bias errors at class-level how to detect and improve individual fairness of a model interpretability, performance, and generalizability in bias testing expose confusion  ... 
arXiv:2205.08809v1 fatcat:63whhiyjvvaida4kjvpsyd7gh4

Fair Dynamic Rationing [article]

Vahideh Manshadi, Rad Niazadeh, Scott Rodilitz
2022 arXiv   pre-print
achieves matching lower bounds for both objectives (ex-post and ex-ante fairness), for any set of parameters.  ...  For an arbitrarily correlated sequence of demands, we establish upper bounds on the expected minimum fill rate (ex-post fairness) and the minimum expected fill rate (ex-ante fairness) achievable by any  ...  Acknowledgment The authors would like to thank Itai Ashlagi, Amin Saberi, and Ed Kaplan for helpful comments and insights at early stages of this work.  ... 
arXiv:2102.01240v4 fatcat:4uyqliwwi5dr3fftoiybcobbwm

A metric characterization of fair computations in CCS [chapter]

Gerardo Costa
1985 Lecture Notes in Computer Science  
(FD,ds), yields the space of all finite derivations together with all -and only -the weakly, resp. strongly, fair computations (i.e. nonextendable derivations).  ...  Then the metric completion of (FD,da) yields the space of all derivations, while the completion of (FD,dw), resp.  ...  More recently, De Bakker and Zucker have proposed metric spaces of processes where they can interpret fair parallel operators using the standard notion of limit of (countable) sequences [BZ] .  ... 
doi:10.1007/3-540-15198-2_15 fatcat:sdigimasr5b2jc3jawunrhosxm

A minimax framework for quantifying risk-fairness trade-off in regression [article]

Evgenii Chzhen, Nicolas Schreuder
2022 arXiv   pre-print
We provide a general post-processing strategy which enjoys fairness, risk guarantees and can be applied on top of any black-box algorithm.  ...  For α∈ (0, 1) the proposed framework allows to continuously interpolate between these two extreme cases and to study partially fair predictors.  ...  This abstract result enables us to characterize explicitly oracle α-RI {f * α } α∈ [0, 1] .  ... 
arXiv:2007.14265v3 fatcat:sy47uva475dolctn2gjylmtpxq

Fairness in Ranking: A Survey [article]

Meike Zehlike, Ke Yang, Julia Stoyanovich
2021 arXiv   pre-print
In the past few years, there has been much work on incorporating fairness requirements into algorithmic rankers, with contributions coming from the data management, algorithms, information retrieval, and  ...  An important contribution of our work is in developing a common narrative around the value frameworks that motivate specific fairness-enhancing interventions in ranking.  ...  The oracle is deterministic, and no further assumptions are made about the type of fairness criteria it encodes.  ... 
arXiv:2103.14000v2 fatcat:6xblra4wtndfdoha5fkd6epzma

Metric Learning for Individual Fairness [article]

Christina Ilvento
2020 arXiv   pre-print
In this work, we propose a solution to the problem of approximating a metric for Individual Fairness based on human judgments.  ...  individuals, and learning procedures to construct hypotheses for metric approximations which generalize to unseen samples under certain assumptions of learnability of distance threshold functions.  ...  [15] and [12] consider the problem of generalizing Individual Fairness with differing levels of oracle access to the metric, and one could view our results as providing a path for efficiently generating  ... 
arXiv:1906.00250v2 fatcat:vz6aw6suwrg5bp2bpkgj52d6h4

Optimized Score Transformation for Consistent Fair Classification [article]

Dennis Wei, Karthikeyan Natesan Ramamurthy, Flavio du Pin Calmon
2021 arXiv   pre-print
metrics such as accuracy.  ...  Comprehensive experiments comparing to 10 existing methods show that FairScoreTransformer has advantages for score-based metrics such as Brier score and AUC while remaining competitive for binary label-based  ...  Metrics Classification performance and fairness were evaluated using both score-based metrics (log loss, Brier score, and AUC for performance, differences in mean scores (MSP) and GEO for fairness) and  ... 
arXiv:1906.00066v3 fatcat:q7apfyik25h6zgev5kb6pcb5ai

Consider the Alternatives: Navigating Fairness-Accuracy Tradeoffs via Disqualification [article]

Guy N. Rothblum, Gal Yona
2021 arXiv   pre-print
The notion of "too much" is quantified via a parameter γ that serves as a vehicle for specifying acceptable tradeoffs between accuracy and fairness, in a way that is independent from the specific metrics  ...  used to quantify fairness and accuracy in a given task.  ...  This difficulty is further exacerbated by the fact that both fairness and accuracy are in general two abstract concepts, that can be operationalized in different ways (e.g. by choosing different metrics  ... 
arXiv:2110.00813v1 fatcat:dyx7do34wjckvebumenxpftr2q

A Subscription Service for Automated Communication and Fair Cost Distribution in Collaborative Blockchainbased Business Processes [chapter]

Moritz Schindelmann, Friedrich-Alexander-Universität Erlangen-Nürnberg, Lehrstuhl für Wirtschaftsinformatik, insb. im Dienstleistungsbereich, Nürnberg, Germany, Philipp Klinger, Freimut Bodendorf
2020 WI2020 Zentrale Tracks  
Evaluation of the prototype and architecture shows that this does not only avoid unnecessary latency in process communication but also results in a fair cost distribution as each participant is only charged  ...  Blockchain capabilities like Ethereum Smart Contracts offer great opportunities to manage cross-organizational business processes due to their trustless and tamperproof nature.  ...  The service is evaluated by simulating an exemplary business process, tracking the metrics process latency and cost distribution, and comparing the results to those of an alternative approach for automating  ... 
doi:10.30844/wi_2020_r13-schindelmann dblp:conf/wirtschaftsinformatik/SchindelmannKB20 fatcat:wap7li7afnfsdktiwjhwysqwfi

SimCLS: A Simple Framework for Contrastive Learning of Abstractive Summarization [article]

Yixin Liu, Pengfei Liu
2021 arXiv   pre-print
In this paper, we present a conceptually simple while empirically powerful framework for abstractive summarization, SimCLS, which can bridge the gap between the learning objective and evaluation metrics  ...  We have open-sourced our codes and results: https://github.com/yixinL7/SimCLS.  ...  The views and conclusions contained herein are those of the authors and should not be interpreted as necessarily representing the official policies or endorsements, either expressed or implied, of the  ... 
arXiv:2106.01890v1 fatcat:735jnyppezh7zlycow7wd7zffi
« Previous Showing results 1 — 15 out of 2,883 results