Privacy preserving collaborative learning of generalized linear mixed model

Md. Monowar Anjum, Noman Mohammed, Wentao Li, Xiaoqian Jiang

Abstract: Highlights•We propose a novel privacy-preserving training method (value-blind parameter optimization) for GLMM in CL setting.•Our proposed method can perform parameter optimization on homomorphically encrypted values shared by the collaborating parties.•We perform extensive experimentation on our proposed training method. Experimental results show that our proposed method achieves low error rate (less than 5%) in both real world and synthetic datasets.