Poisoning Attacks against Support Vector Machines
arXiv:1206.6389
Abstract
We investigate a family of poisoning attacks against Support Vector Machines (SVM). Such attacks inject specially crafted training data that increases the SVM's test error. Central to the motivation for these attacks is the fact that most learning algorithms assume that their training data comes from a natural or well-behaved distribution. However, this assumption does not generally hold in security-sensitive settings. As we demonstrate, an intelligent adversary can, to some extent, predict the change of the SVM's decision function due to malicious input and use this ability to construct malicious data. The proposed attack uses a gradient ascent strategy in which the gradient is computed based on properties of the SVM's optimal solution. This method can be kernelized and enables the attack to be constructed in the input space even for non-linear kernels. We experimentally demonstrate that our gradient ascent procedure reliably identifies good local maxima of the non-convex validation error surface, which significantly increases the classifier's test error.
Appears in Proceedings of the 29th International Conference on Machine Learning (ICML 2012)
Cited by in corpus (11)
- Security Evaluation of Pattern Classifiers under Attack
- Terminal Brain Damage: Exposing the Graceless Degradation in Deep Neural Networks Under Hardware Fault Attacks
- Adversarial Samples on Android Malware Detection Systems for IoT Systems
- Randomized Prediction Games for Adversarial Machine Learning
- Effectiveness of Distillation Attack and Countermeasure on Neural Network Watermarking
- A Formalization of Robustness for Deep Neural Networks
- Learning to Confuse: Generating Training Time Adversarial Data with Auto-Encoder
- Lower Bounds for Adversarially Robust PAC Learning
- AIR5: Five Pillars of Artificial Intelligence Research
- Investigating Robustness and Interpretability of Link Prediction via Adversarial Modifications
- Adequacy of the Gradient-Descent Method for Classifier Evasion Attacks