Attacks on Online Learners: a Teacher-Student Analysis
On the map: Knowledge distillation → Federated, robust and private distillation
Abstract (arXiv)
Machine learning models are famously vulnerable to adversarial attacks: small ad-hoc perturbations of the data that can catastrophically alter the model predictions. While a large literature has studied the case of test-time attacks on pre-trained models, the important case of attacks in an online learning setting has received little attention so far. In this work, we use a control-theoretical perspective to study the scenario where an attacker may perturb data labels to manipulate the learning dynamics of an online learner. We perform a theoretical analysis of the problem in a teacher-student setup, considering different attack strategies, and obtaining analytical results for the steady state of simple linear learners. These results enable us to prove that a discontinuous transition in the learner's accuracy occurs when the attack strength exceeds a critical threshold. We then study empirically attacks on learners with complex architectures using real data, confirming the insights of our theoretical analysis. Our findings show that greedy attacks can be extremely efficient, especially when data stream in small batches.
Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in Federated, robust and private distillation, one of 12 directions in Knowledge distillation. Builds on 18 of its 33 references that are in the map. First posted on arXiv 18 May 2023.
Builds on
18 of 33 references.
Closest ancestors · 1
Same direction, newest first
Other references · 17
- 2015Deep Residual Learning for Image Recognitioncited by 52,096
- 2014Very Deep Convolutional Networks for Large-Scale Image Recognitioncited by 15,364
- 2015Continuous control with deep reinforcement learningcited by 3,407
- 2018Addressing Function Approximation Error in Actor-Critic Methodscited by 2,417
- 2017Targeted Backdoor Attacks on Deep Learning Systems Using Data Poisoningcited by 1,055
- 2018Robustness May Be at Odds with Accuracycited by 880
- 2017Wild Patterns: Ten Years After the Rise of Adversarial Machine Learningcited by 414
- 2017Towards Poisoning of Deep Learning Algorithms with Back-gradient Optimizationcited by 197
- 2020Dataset Security for Machine Learning: Data Poisoning, Backdoor Attacks, and Defensescited by 180
- 2021Continual Learning in the Teacher-Student Setup: Impact of Task Similaritycited by 88
- 2019Policy Poisoning in Batch Reinforcement Learning and Controlcited by 70
- 2020Robustness of Bayesian Neural Networks to Gradient-Based Attackscited by 48
5 more
- 2017Iterative Machine Teachingcited by 41
- 2018Data Poisoning Attacks against Online Learningcited by 27
- 2022Support Vector Machines under Adversarial Label Contaminationcited by 17
- 2018An Optimal Control Approach to Sequential Machine Teachingcited by 11
- 2021Accumulative Poisoning Attacks on Real-time Datacited by 10
Built on by
Not cited yet.
Similar, not cited
Close in wording, no citation either way
- 2025Keeping up with dynamic attackers: Certifying robustness to adaptive online data poisoning
- 2021Influence Based Defense Against Data Poisoning Attacks in Online Learning
- 2022Wild Patterns Reloaded: A Survey of Machine Learning Security against Training Data Poisoning
- 2023What Distributions are Robust to Indiscriminate Poisoning Attacks for Linear Learners?
- 2026Evaluating Answer Leakage Robustness of LLM Tutors against Adversarial Student Attacks
- 2024On the Adversarial Risk of Test Time Adaptation: An Investigation into Realistic Test-Time Data Poisoning
- 2021Reliable Adversarial Distillation with Unreliable Teachers
- 2025How to Backdoor Image Knowledge Distillation