Secure Bilevel Asynchronous Vertical Federated Learning with Backward Updating

Vertical federated learning (VFL) attracts increasing attention due to the emerging demands of multi-party collaborative modeling and concerns privacy leakage. In real VFL applications, usually only one or partial parties hold labels, which makes it challenging for all collaboratively learn model without Meanwhile, most existing algorithms are trapped in synchronous computations, leads inefficiency their real-world applications. To address these problems, we propose a novel framework integrated with new backward updating mechanism bilevel asynchronous parallel architecture (VFB^2), under three algorithms, including VFB^2-SGD, -SVRG, -SAGA, proposed. We derive theoretical results convergence rates both strongly convex nonconvex conditions. also prove security VFB^2 semi-honest threat models. Extensive experiments on benchmark datasets demonstrate that our efficient, scalable, lossless.