DLFTI: A deep learning based fast truth inference mechanism for distributed spatiotemporal data in mobile crowd sensing

Jianheng Tang, Kejia Fan, Pengzhi Yin, Zhenzhe Qu, Anfeng Liu, Neal N. Xiong, Tian Wang, Mianxiong Dong, Shaobo Zhang

Abstract: Highlights•A Deep Learning based Fast Truth Inference (DLFTI) mechanism is proposed to accomplish fast and accurate truth inference in MCS, which is resistant to fraudulent attack by malicious gangs.•A multi-level Ground Truth Data (GTD) and multi-level Estimated True Data (ETD) framework is proposed for fast trust computing and accurate truth discovery.•A worker recognition method is proposed to quickly recognize workers by establishing worker profiles based on the workers’ degrees-of-trust.•Deep Matrix Factorization algorithm are adopted to leverage of the spatiotemporal correlation between the distributed data.•Experimental results show that the proposed mechanism can effectively perform truth inference in a wide range, even in the extreme cases where malicious workers are in the majority.