Adaptive Federated Learning with Differential Privacy Constraints
Abstract
Federated learning (FL) offers a promising approach to training machine learning models on decentralized data sources without directly exchanging the data itself. However, the inherent heterogeneity of data distributions across clients and the need to protect user privacy pose significant challenges. This paper proposes a novel adaptive federated learning algorithm that dynamically adjusts both the learning rate and differential privacy parameters based on client-specific data characteristics. The algorithm incorporates a real-time feedback loop to monitor model performance, allowing for adaptive adjustments aimed at accelerating convergence and minimizing information leakage. We demonstrate the effectiveness of this approach through a theoretical analysis and discuss its potential for improving the efficiency and security of federated learning in diverse scenarios. The core of the algorithm lies in a dynamic adjustment mechanism that considers client data heterogeneity and privacy constraints, resulting in a more robust and personalized learning experience. The proposed method tackles the limitations of traditional FL algorithms by providing a mechanism to handle data imbalances and simultaneously preserve 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.