• Publications
  • Influence
Equality of Opportunity in Supervised Learning
TLDR
We propose a criterion for discrimination against a specified sensitive attribute in supervised learning, where the goal is to predict some target based on available features. Expand
Understanding deep learning requires rethinking generalization
TLDR
We show that state-of-the-art convolutional networks for image classification trained with stochastic gradient methods easily fit a random labeling of the training data. Expand
Fairness through awareness
TLDR
We study fairness in classification, where the goal is to prevent discrimination against individuals based on their membership in some group, while maintaining utility for the classifier (the university). Expand
Train faster, generalize better: Stability of stochastic gradient descent
TLDR
We show that parametric models trained by a stochastic gradient method (SGM) with few iterations have vanishing generalization error. Expand
Sanity Checks for Saliency Maps
TLDR
We propose an actionable methodology based on randomization tests to evaluate the adequacy of explanation approaches. Expand
A Simple and Practical Algorithm for Differentially Private Data Release
TLDR
We present a new algorithm for differentially private data release, based on a simple combination of the Multiplicative Weights update rule with the Exponential Mechanism. Expand
On the geometry of differential privacy
TLDR
We consider the noise complexity of differentially private mechanisms in the setting where the user asks d linear queries f:Rn -> R non-adaptively. Expand
Preserving Statistical Validity in Adaptive Data Analysis
TLDR
We propose a principled study of how to guarantee the validity of statistical inference in adaptive data analysis. Expand
A Multiplicative Weights Mechanism for Privacy-Preserving Data Analysis
TLDR
We propose a new differentially private multiplicative weights mechanism for answering a large number of interactive counting (or linear) queries that arrive online and may be adaptively chosen. Expand
Avoiding Discrimination through Causal Reasoning
TLDR
We frame the problem of discrimination based on protected attributes in the language of causal reasoning and propose natural causal non-discrimination criteria. Expand
...
1
2
3
4
5
...