Reproduce “Vertical Federated Learning: Concepts, Advances and Challenges” 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
Yang Liu, Yan Kang, Tianyuan Zou, Yanhong Pu, Yuanqin He, Xiaozhou Ye, Ye Ouyang, Ya-Qin Zhang and 1 more
This comprehensive survey consolidates the concepts, advances, and challenges in Vertical Federated Learning (VFL), proposing a novel VFLow framework for constrained multi-objective optimization. It systematically categorizes VFL architectures, efficiency/effectiveness techniques, privacy-preserving protocols, and attack/defense strategies, guiding future research and development.
Key insights
- The VFLow framework provides a holistic perspective for VFL algorithm design, advocating for the simultaneous consideration of effectiveness, efficiency, privacy, and fairness as intertwined constraints.
- A structured classification of VFL architectures (splitVFL, aggVFL, splitVFLc, aggVFLc) and privacy protocols offers a standardized vocabulary for discussing and evaluating VFL systems.
- Identification of crucial open challenges—including interoperability, trustworthy VFL, automated VFL, data valuation, fairness, and the need for diverse datasets—provides a clear roadmap for future research directions.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.