Privacy-Friendly IoT: Does federated learning guarantee privacy?
Abstract
Abstract Federated learning is an emerging machine learning paradigm where multiple clients train models locally and formulate a global model based on the local model updates. Privacy is one of its essential properties. To determine if federated learning guarantees privacy we study security threats that jeopardize privacy. We also look at these security threats in terms of federated learning applications in the internet of things. Two of these applications include data sharing and giving feedback system. It seems that in the IoT network, some security threats such as deep leakage from gradients and inverting gradients can be eliminated using methods based on differential privacy, but others, such as neural backdoor, are still effective and there is no guarantee of privacy.
Read the paper
Related studies
- Active Learning during Federated Learning for Object Detection
- Fairness-aware federated learning
- Decentralized Federated Learning
- Building a Personally Identifiable Information Recognizer in a Privacy Preserved Manner Using Automated Annotation and Federated Learning
- Preservation of Patient Level Privacy: Federated Classification and Calibration Models
- Efficient explainable federated learning
Go deeper in Evidence Hub
Evidence Hub adds a study data sheet (design, sample size, evidence level, key numbers), an AI summary and a chat grounded in the paper's text.