Limitations of statistical and causal notions of fairnessChris Hays
How the definitions of social groups matter for mathematically formalizing fairness.
Big data "summaries" can speed up regression on panel data — and reduce the risks of data breachesChris Hays
Coresets can efficiently represent data for common statistics and machine learning algorithms; a new paper shows how to construct them for regressions on panel data.
Disentangling the market dynamics of GDPR, EU’s data privacy legislationChris Hays
A novel model of the GDPR in personal data driven markets nuances the debate about business interests and consumer privacy.
Debiasing data: An efficient framework using the principle of maximum entropyChris Hays
An old idea from statistical physics leads to a new algorithm for mitigating societal biases in data.