Published on Thu Oct 24 2019

Preventing Adversarial Use of Datasets through Fair Core-Set Construction

Benjamin Spector, Ravi Kumar, Andrew Tomkins

The core-set allows strong performance on primary tasks, but forcespoor performance on unwanted tasks. We give methods for both linear models and neural networks and demonstrate their efficacy on data.

0
0
0
Abstract

We propose improving the privacy properties of a dataset by publishing only a strategically chosen "core-set" of the data containing a subset of the instances. The core-set allows strong performance on primary tasks, but forces poor performance on unwanted tasks. We give methods for both linear models and neural networks and demonstrate their efficacy on data.

Thu Oct 04 2018
Machine Learning
Finding Solutions to Generative Adversarial Privacy
We present heuristics for solving the maximin problem induced by the generative adversarial privacy setting for linear and convolutional neural network (CNN) adversaries. In the linear adversary setting, we present a greedy algorithm for approximating the optimal solution for the privatizer.
0
0
0
Mon Jul 16 2018
Machine Learning
Machine Learning with Membership Privacy using Adversarial Regularization
Machine learning models leak information about the datasets on which they are trained. An adversary can build an algorithm to trace the individual members of a model's training dataset. We introduce a privacy mechanism to train machine learning models that achieve membership privacy.
0
0
0
Tue Jul 25 2017
Machine Learning
Proxy Non-Discrimination in Data-Driven Systems
Machine learnt systems inherit biases against protected classes, historically disparaged groups, from training data. Usually, these biases are not explicit, rely on subtle correlations discovered by training algorithms, and are difficult to detect. We formalize proxy discrimination as a class of properties indicative of bias.
0
0
0
Fri Dec 07 2018
Machine Learning
Three Tools for Practical Differential Privacy
Differentially private learning on real-world data poses challenges for standard machine learning practice. Privacy guarantees are difficult to interpret, hyperparameter tuning on private data reduces the privacy budget. Ad-hoc privacy attacks are often required to test model privacy.
0
0
0
Fri Feb 09 2018
Machine Learning
Certified Robustness to Adversarial Examples with Differential Privacy
Adversarial examples that fool machine learning models have been a topic of intense research interest. Most past defenses are vulnerable to sophisticated attacks. This paper presents the first certified defense that both scales to large networks and datasets.
1
0
0
Mon Jun 07 2021
Machine Learning
Antipodes of Label Differential Privacy: PATE and ALIBI
0
0
0