• Publications
  • Influence
Making Deep Neural Networks Robust to Label Noise: A Loss Correction Approach
TLDR
It is proved that, when ReLU is the only non-linearity, the loss curvature is immune to class-dependent label noise, and it is shown how one can estimate these probabilities, adapting a recent technique for noise estimation to the multi-class setting, and providing an end-to-end framework.
Private federated learning on vertically partitioned data via entity resolution and additively homomorphic encryption
TLDR
This work describes a three-party end-to-end solution in two phases ---privacy-preserving entity resolution and federated logistic regression over messages encrypted with an additively homomorphic scheme---, secure against a honest-but-curious adversary.
Making Neural Networks Robust to Label Noise: a Loss Correction Approach
TLDR
It is proved that, when ReLU is the only non-linearity, the loss curvature is immune to class-dependent label noise, and the proposed procedures for loss correction simply amount to at most a matrix inversion and multiplication.
Loss factorization, weakly supervised learning and label noise robustness
We prove that the empirical risk of most well-known loss functions factors into a linear term aggregating all labels with a term that is label free, and can further be expressed by sums of the loss.
(Almost) No Label No Cry
TLDR
This work shows that the mean operator, a statistic which aggregates all labels, is minimally sufficient for the minimization of many proper scoring losses with linear (or kernelized) classifiers without using labels, and provides a fast learning algorithm that estimates the mean operators via a manifold regularizer with guaranteed approximation bounds.
Tsallis Regularized Optimal Transport and Ecological Inference
TLDR
The first application of optimal transport to the problem of ecological inference, that is, the reconstruction of joint distributions from their marginals, is presented, a problem of large interest in the social sciences.
Entity Resolution and Federated Learning get a Federated Resolution
TLDR
This paper provides a thorough answer to how optimal classifiers, empirical losses, margins and generalisation abilities are affected by entity resolution, and brings simple practical arguments to upgrade entity resolution as a preprocessing step to learning.
Sinkhorn AutoEncoders
TLDR
It is proved that optimizing the encoder over any class of universal approximators, such as deterministic neural networks, is enough to come arbitrarily close to the optimum, and advertised this framework, which holds for any metric space and prior, as a sweet-spot of current generative autoencoding objectives.
Three Tools for Practical Differential Privacy
TLDR
Three tools to make differentially private machine learning more practical are introduced: simple sanity checks which can be carried out in a centralized manner before training, an adaptive clipping bound to reduce the effective number of tuneable privacy parameters, and it is shown that large-batch training improves model performance.
SEALion: a Framework for Neural Network Inference on Encrypted Data
We present SEALion: an extensible framework for privacy-preserving machine learning with homomorphic encryption. It allows one to learn deep neural networks that can be seamlessly utilized for
...
...