Postdoc, City University of Hong Kong
1 paper at NeurIPS 2025
We propose PubSub-VFL, a novel VFL paradigm with a Publisher/Subscriber architecture optimized for two-party collaborative learning with high computational efficiency.