Peer-to-peer federated learning
Abstract
Federated Learning is a transformative approach for decentralized machine learning, enabling collaborative model training without centralized data aggregation. This thesis investigates a more decentralized variant of it, Peer-to-Peer Federated Learning, which emphasizes peer autonomy, communication efficiency, and privacy preservation in distributed, often resource-constrained environments. By exploring its foundational concepts and practical implementations, this work positions Peer-to-Peer Federated Learning as a promising direction within the evolving landscape of decentralized AI. A major contribution of this work is PeerFL, a modular simulation platform that enables realistic experimentation of Federated Learning with communication protocols, adversarial settings, and resource constraints. To tackle scalability and efficiency, this work explores algorithmic strategies such as probabilistic device selection, topology optimization via graph algorithms, and clustering based on game theory. Privacy and security challenges in this setting are also analyzed in depth. Another major contribution of this work is a proposal for a Federated Learning Marketplace, an open platform for secure model sharing, incentivization, and legal compliance. This was implemented through a cloud-based backend and Android app, offering practical insights into real-world deployment. Practical applications of Federated Learning to LLM-backed medical chatbots, recommendation systems and deepfake…
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.