长短时记忆网络在移动场景中的应用研究进展  

Research Progress of Long Short-Term Memory Applications in Mobile Scenarios

作　　者：孙运淼 林锋[1] 周激流[1] SUN Yun-miao;LIN Feng;ZHOU Ji-liu(College of Computer Science, Sichuan University, Chengdu 610065)

机构地区：[1]四川大学计算机学院,成都610065

出　　处：《现代计算机（中旬刊）》2017年第12期10-15,共6页Modern Computer

基　　金：国家自然科学基金(No.61272448);高等学校博士学科点专项科研基金(No.20110181130007);川大-泸州战略合作项目(No.20826041A4345)

摘　　要：移动定位技术的普及使得海量的移动轨迹得到采集与存储,随之也带来轨迹数据处理的巨大需求和研究价值。长短时记忆网络因其出色的序列数据处理能力,近年来已经在该领域得到许多关注与应用。对长短时记忆网络的起源背景进行回顾,介绍其工作机制以及发展现状,并着重探讨在移动场景中,利用长短时记忆网络实现轨迹数据处理的几种典型应用。A massive volume of trajectory data has been collected and stored since the popularization of Location Based Services, hence the huge re- quirement and research value of trajectory data processing. Long Short-Term Memory （LSTM） has won lots of attention and applications due to its excellent ability of sequence data processing. Reviews the research background of LSTM, introduces its mechanism and current development, and elaborates several typical applications of LSTM in trajectory data processing in mobile scenarios.

关 键 词：长短时记忆网络 循环神经网络 移动场景 轨迹数据处理 时间序列分析 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]