Online Decentralized Frank-Wolfe: From theoretical bound to applications in smart-building
This addresses efficient distributed learning for applications like smart buildings, but appears incremental as it builds on Frank-Wolfe methods.
The authors tackled the problem of decentralized learning with non-convex loss functions in networks with limited resources, proposing an online algorithm and providing theoretical guarantees, with demonstration on a smart-building application.
The design of decentralized learning algorithms is important in the fast-growing world in which data are distributed over participants with limited local computation resources and communication. In this direction, we propose an online algorithm minimizing non-convex loss functions aggregated from individual data/models distributed over a network. We provide the theoretical performance guarantee of our algorithm and demonstrate its utility on a real life smart building.