Reproduce “MetaFed: Advancing Privacy, Performance, and Sustainability in Federated Metaverse Systems” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Muhammet Anil Yagiz, Zeynep Sude Cengiz, Polat Goktas
The MetaFed framework introduces a decentralized federated learning system for Metaverse environments, concurrently optimizing for performance, privacy, and sustainability. It achieves up to 80.26% accuracy on CIFAR-10 and reduces per-round CO2 emissions by nearly 50% compared to traditional federated learning methods.
Key insights
- Effective multi-objective optimization for federated systems can be achieved by synergistically integrating intelligent orchestration, privacy-enhancing technologies, and sustainability-aware mechanisms.
- MARL-based resource orchestration significantly improves model accuracy and convergence stability, particularly for more complex datasets and heterogeneous data distributions.
- Carbon-aware client scheduling, guided by real-time energy grid data, can substantially reduce the environmental footprint of federated learning without compromising predictive performance.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.