2018YFE0126000-005-基于深度学习的车联网交通流预测测试数据

针对涉及多方参与的传统推荐流程存在用户私密信息易被复制或盗取等安全威胁的问题，利用深度学习提供个性化推荐服务隐私保护方案，在保护用户隐私安全的前提下，准确、高效地向用户提供更方便的位置推荐服务。基于深度学习的车联网交通流预测测试数据集主要利用美国明尼苏达大学交通数据研究实验室的双子城交通数据对该框架进行实验分析，记录了对公开数据进行特征处理后的数据、剔除异常节点后的数据、路网数据平稳性、压缩后的路网数据、数据浅层的序列特征、自编码数据、数据深层的隐藏特征、预测精度等。数据量大小26.2M。

Aiming at the security threats such as easy replication and theft of users' private information in the traditional recommendation process involving multiple parties, this study proposes a privacy-preserving scheme for personalized recommendation services using deep learning, which delivers accurate, efficient and more convenient location recommendation services to users while protecting their privacy and security. The test dataset for deep learning-based Internet of Vehicles (IoV) traffic flow prediction uses the Twin Cities traffic data from the Traffic Data Research Laboratory of the University of Minnesota, USA for experimental validation of the proposed framework. The dataset records the data after feature processing of public data, data after removing abnormal nodes, stationarity of road network data, compressed road network data, shallow sequential features of the data, auto-encoded data, deep hidden features of the data, prediction accuracy, and other related contents. The total size of the dataset is 26.2M.

搜集汇总

背景与挑战

背景概述

该数据集是针对车联网交通流预测的测试数据，利用深度学习技术提供隐私保护方案，以应对传统推荐流程中的安全威胁。它基于美国明尼苏达大学双子城交通数据，包含特征处理、路网数据、序列特征、自编码数据及预测精度等内容，数据量约26.2M，共65个文件。