基于深度学习的分布式时空数据快速真值推理方法

A Fast Truth Reasoning Method for Distributed Spatiotemporal Data Based on Deep Learning

针对移动群众分布式感知中真值推理缺乏有效的真值基准数据和参与者可信度信息的问题,提出了基于深度学习的分布式时空数据快速真值推理方法(DLFTI),构建了金牌真值数据(来自无人机)、银牌真值数据(来自高可信参与者)和铜牌真值数据(来自深度矩阵分解)三级真值基准数据框架,实现了对参与者的快速可信度计算,同时设计了三级估计真值数据并成功实现了移动群众感知的快速且精确的真值推理。仿真实验结果表明:本文方法与传统算法相比,在参与者识别和真值发现方面,性能得到显著提升。

In response to the lack of effective ground truth benchmarks and participant credibility information in distributed sensing of mobile crowds,this paper introduces a Deep Learning-based Fast Truth Inference method for distributed spatiotemporal data(DLFTI).It establishes a three-tiered ground truth benchmark framework consisting of gold truth data(sourced from drones),silver truth data(sourced from highly credible participants),and bronze truth data(derived from deep matrix factorization),enabling rapid credibility calculation of participants.Additionally,the method is designed to estimate truth data across three levels and has successfully achieved fast and accurate truth inference in mobile crowd sensing.Simulation results indicate that this method significantly outperforms traditional algorithms in participant identification and truth discovery.