A Survey on Vulnerability of Federated Learning: A Learning Algorithm Perspective
arXiv:2311.16065 · doi:10.1016/j.neucom.2023.127225
Abstract
This review paper takes a comprehensive look at malicious attacks against FL, categorizing them from new perspectives on attack origins and targets, and providing insights into their methodology and impact. In this survey, we focus on threat models targeting the learning process of FL systems. Based on the source and target of the attack, we categorize existing threat models into four types, Data to Model (D2M), Model to Data (M2D), Model to Model (M2M) and composite attacks. For each attack type, we discuss the defense strategies proposed, highlighting their effectiveness, assumptions and potential areas for improvement. Defense strategies have evolved from using a singular metric to excluding malicious clients, to employing a multifaceted approach examining client models at various phases. In this survey paper, our research indicates that the to-learn data, the learning gradients, and the learned model at different stages all can be manipulated to initiate malicious attacks that range from undermining model performance, reconstructing private local data, and to inserting backdoors. We have also seen these threat are becoming more insidious. While earlier studies typically amplified malicious gradients, recent endeavors subtly alter the least significant weights in local models to bypass defense measures. This literature review provides a holistic understanding of the current FL threat landscape and highlights the importance of developing robust, efficient, and privacy-preserving defenses to ensure the safe and trusted adoption of FL in real-world applications.
arXiv commenthttps://github.com/Rand2AI/Awesome-Vulnerability-of-Federated-Learning
References in corpus (32)
- Very Deep Convolutional Networks for Large-Scale Image Recognition
- Fashion-MNIST: a Novel Image Dataset for Benchmarking Machine Learning Algorithms
- Score-Based Generative Modeling through Stochastic Differential Equations
- On the Convergence of FedAvg on Non-IID Data
- BadNets: Identifying Vulnerabilities in the Machine Learning Model Supply Chain
- Learning Differentially Private Recurrent Language Models
- The Hidden Vulnerability of Distributed Learning in Byzantium
- iDLG: Improved Deep Leakage from Gradients
- Can You Really Backdoor Federated Learning?
- Mitigating Sybils in Federated Learning Poisoning
- Threats to Federated Learning: A Survey
- Federated Collaborative Filtering for Privacy-Preserving Personalized Recommendation System
- Federated Social Recommendation with Graph Neural Network
- DeepSight: Mitigating Backdoor Attacks in Federated Learning Through Deep Model Inspection
- Generalized Byzantine-tolerant SGD
- Privacy Preserving Distributed Machine Learning with Federated Learning
- Distributed Statistical Machine Learning in Adversarial Settings: Byzantine Gradient Descent
- Neural Attention Distillation: Erasing Backdoor Triggers from Deep Neural Networks
- GRNN: Generative Regression Neural Network -- A Data Leakage Attack for Federated Learning
- Efficient privacy preservation of big data for accurate data mining
- Efficient Parallel Methods for Deep Reinforcement Learning
- Efficient Data Perturbation for Privacy Preserving and Accurate Data Stream Mining
- Free-riders in Federated Learning: Attacks and Defenses
- Mitigating Backdoor Attacks in Federated Learning
- Poisoning Deep Learning Based Recommender Model in Federated Learning Scenarios
- Federated Learning Meets Natural Language Processing: A Survey
- Turning Federated Learning Systems Into Covert Channels
- Relational Multi-Task Learning: Modeling Relations between Data and Tasks
- Thinking Two Moves Ahead: Anticipating Other Users Improves Backdoor Attacks in Federated Learning
- Learning to Backdoor Federated Learning
- Local Environment Poisoning Attacks on Federated Reinforcement Learning
- Gradient Leakage Defense with Key-Lock Module for Federated Learning