Private federated learning on vertically partitioned data via entity resolution and additively homomorphic encryption
Consider two data providers, each maintaining private records of different feature sets about
common entities. They aim to learn a linear model jointly in a federated setting, namely, data …
common entities. They aim to learn a linear model jointly in a federated setting, namely, data …
Adaboost-LLP: A boosting method for learning with label proportions
How to solve the classification problem with only label proportions has recently drawn
increasing attention in the machine learning field. In this paper, we propose an ensemble …
increasing attention in the machine learning field. In this paper, we propose an ensemble …
A theoretical framework for robustness of (deep) classifiers against adversarial examples
Most machine learning classifiers, including deep neural networks, are vulnerable to
adversarial examples. Such inputs are typically generated by adding small but purposeful …
adversarial examples. Such inputs are typically generated by adding small but purposeful …
Privacy-preserving class ratio estimation
In this paper we present learning models for the class ratio estimation problem, which takes
as input an unlabeled set of instances and predicts the proportions of instances in the set …
as input an unlabeled set of instances and predicts the proportions of instances in the set …
Fast learning from distributed datasets without entity matching
Consider the following data fusion scenario: two datasets/peers contain the same real-world
entities described using partially shared features, eg banking and insurance company …
entities described using partially shared features, eg banking and insurance company …
[PDF][PDF] Weakly supervised learning via statistical sufficiency
G Patrini - 2016 - core.ac.uk
The Thesis introduces a novel algorithmic framework for weakly supervised learning,
namely, for any any problem in between supervised and unsupervised learning, from the …
namely, for any any problem in between supervised and unsupervised learning, from the …
On regularizing rademacher observation losses
R Nock - Advances in Neural Information Processing …, 2016 - proceedings.neurips.cc
It has recently been shown that supervised learning linear classifiers with two of the most
popular losses, the logistic and square loss, is equivalent to optimizing an equivalent loss …
popular losses, the logistic and square loss, is equivalent to optimizing an equivalent loss …
Learning games and Rademacher observations losses
R Nock - arXiv preprint arXiv:1512.05244, 2015 - arxiv.org
It has recently been shown that supervised learning with the popular logistic loss is
equivalent to optimizing the exponential loss over sufficient statistics about the class …
equivalent to optimizing the exponential loss over sufficient statistics about the class …
[PDF][PDF] On Private Supervised Distributed Learning: Weakly Labeled and without Entity Resolution
We describe a system with strong privacy guarantees that is able to learn (supervised) linear
classifiers in the challenging setting where data is distributed, entity matching/resolution is …
classifiers in the challenging setting where data is distributed, entity matching/resolution is …
[PDF][PDF] Privacy-preserving entity resolution and logistic regression on encrypted data
G Patrini - giorgiop.github.io
Privacy-preserving entity resolution and logistic regression on encrypted data Page 1
Privacy-preserving entity resolution and logistic regression on encrypted data Giorgio Patrini & …
Privacy-preserving entity resolution and logistic regression on encrypted data Giorgio Patrini & …