Intrinsic Certified Robustness of Bagging against Data Poisoning Attacks
Preprint
Abstract
In a \emph{data poisoning attack}, an attacker modifies, deletes, and/or inserts some training examples to corrupt the learnt machine learning model. \emph{Bootstrap Aggregating (bagging)} is a well-known ensemble learning method, which trains multiple base models on random subsamples of a training dataset using a base learning algorithm and uses majority vote to predict labels of testing examples. We prove the intrinsic certified robustness of bagging against data poisoning attacks. Specifically, we show that bagging with an arbitrary base learning algorithm provably predicts the same label for a testing example when the number of modified, deleted, and/or inserted training examples is bounded by a threshold. Moreover, we show that our derived threshold is tight if no assumptions on the base learning algorithm are made. We evaluate our method on MNIST and CIFAR10. For instance, our method achieves a certified accuracy of $91.1\%$ on MNIST when arbitrarily modifying, deleting, and/or inserting 100 training examples.
Keywords
All Related Versions
This publication has 12 references indexed in Scilit:
- Stronger data poisoning attacks break data sanitization defensesMachine Learning, 2021
- Manipulating Machine Learning: Poisoning Attacks and Countermeasures for Regression Learning2018
- Towards Poisoning of Deep Learning Algorithms with Back-gradient Optimization2017
- Deep Learning with Differential Privacy2016
- Deep Residual Learning for Image Recognition2016
- Poisoning behavioral malware clustering2014
- Systematic Poisoning Attacks on and Defenses for Machine Learning in HealthcareIEEE Journal of Biomedical and Health Informatics, 2014
- Is data clustering in adversarial settings secure?2013
- Casting out Demons: Sanitizing Training Data for Anomaly Sensors2008 IEEE Symposium on Security and Privacy (SP 2008), 2008
- IX. On the problem of the most efficient tests of statistical hypothesesPhilosophical Transactions of the Royal Society A, 1933