基于关系转移和增强学习的时空大数据动态预测

doi:10.6040/j.issn.1672-3961.0.2020.233

山东大学学报 (工学版) ›› 2021, Vol. 51 ›› Issue (2): 105-114.doi: 10.6040/j.issn.1672-3961.0.2020.233

基于关系转移和增强学习的时空大数据动态预测

郑子君^1,2,冯翔^1,2*,虞慧群^1,2,李修全³

1. 华东理工大学计算机科学与工程系, 上海 200237;2. 上海智慧能源工程技术研究中心, 上海 200237;3. 中国科学技术发展战略研究院, 北京 100038

发布日期:2021-04-16
作者简介:郑子君(1994— ),女,江西抚州人,博士研究生,CCF学生会员,主要研究方向为分布并行计算,时空大数据. E-mail:mazjzheng@163.com. *通信作者简介:冯翔(1977— ),女,湖北武汉人,教授,博士,CCF会员,主要研究方向为分布并行计算,计算机网络. E-mail:xfeng@ecust.edu.cn
基金资助:
国家自然科学基金项目(61772200,61772201,61602175);上海市浦江人才计划(17PJ1401900);上海市经信委“信息化发展专项资金”(201602008)

Dynamic prediction of spatiotemporal big data based on relationship transfer and reinforcement learning

ZHENG Zijun^1,2, FENG Xiang^1,2*, YU Huiqun^1,2, LI Xiuquan³

1. Department of Computer Science and Engineering, East China University of Science and Technology, Shanghai 200237, China;
2. Shanghai Engineering Research Center of Smart Energy, Shanghai 200237, China;
3. China Academy of Science and Technology for Development, Beijing 100038, China

Published:2021-04-16

摘要/Abstract

摘要： 为了解决较大时空范围内的动态预测无法获得精确解的问题,采用支持较复杂工作流模式的群智计算方式,提出一种基于关系转移和增强学习的动态预测算法,解决时空数据中的优化问题。设计一个关系转移块,通过对时空数据进行特征提取来学习关系转移概率。建立一个预测增强学习块,随时间序列并行处理转移关系概率,根据特征偏好对时空数据进行优先排序,进而预测问题状态趋势。采用一种深度多步迭代策略优化方法,获得合理的解。从理论上详细地分析和讨论所提出算法的收敛性和收敛速率。在专利转移数据上的试验结果验证了该方法的优势,并证明通过应用关系转移块和预测增强学习块排序精度能得到明显地改善。

关键词: 时空数据, 复杂工作流, 关系转移, 特征学习, 增强学习

Abstract: A dynamic prediction algorithm based on relationship transfer and reinforcement learning was proposed to alleviate the problem that the dynamic prediction in a large spatiotemporal range fails to obtain an accurate solution. The algorithm adopted a crowd intelligence computing manner with complex workflow models to solve the spatiotemporal data optimization problem. A relationship transfer block was designed to learn the probability of relationship transfer by extracting features from spatiotemporal data. A prediction reinforcement learning block was established along with the time series to process the transition relationship probability in parallel and prioritize the spatiotemporal data according to feature preferences that predict the problem status trend. A deep multi-step iterative strategy optimization was adopted to obtain a reasonable solution. Theoretical analysis and discussion of the convergence and convergence rate of the proposed algorithm were conducted. Experimental results on patent transfer data verified this approach's strengths and demonstrated that the ranking accuracy could be significantly improved by applying the relationship transfer block and prediction reinforcement learning block.

Key words: spatiotemporal data, complex workflow, relationship transfer, feature learning, reinforcement learning

中图分类号:

TP391

郑子君,冯翔,虞慧群,李修全. 基于关系转移和增强学习的时空大数据动态预测[J]. 山东大学学报 (工学版), 2021, 51(2): 105-114.

ZHENG Zijun, FENG Xiang, YU Huiqun, LI Xiuquan. Dynamic prediction of spatiotemporal big data based on relationship transfer and reinforcement learning[J]. Journal of Shandong University(Engineering Science), 2021, 51(2): 105-114.

参考文献

[1] LU H, SU S, TIAN Z, et al. A novel search engine for Internet of everything based on dynamic prediction[J]. China Communications, 2019, 16(3): 42-52.
[2] ZHANG H, HU B, WANG X, et al. An action dependent heuristic dynamic programming approach for algal bloom prediction with time-varying parameters[J]. IEEE Access, 2020, 8: 26235-26246.
[3] ROZENSHTEIN P, GIONIS A. Temporal pagerank[C] //Joint European Conference on Machine Learning and Knowledge Discovery in Databases. Cham, Switzerland: Springer, 2016: 674-689.
[4] SUN C, BAI L, KANG L, et al. An approach for predicting uncertain spatiotemporal XML data integrated with grey dynamic model[J]. IEEE Access, 2018, 6: 46801-46825.
[5] MOHAMMADI M, Al-FUQAHA A. Exploiting the spatio-temporal patterns in IoT data to establish a dynamic ensemble of distributed learners[J]. IEEE Access, 2018, 6: 63316-63328.
[6] PAGE L, BRIN S, MOTWANI R, et al. The PageRank citation ranking: Bringing order to the web[R]. Stanford, USA: Stanford InfoLab, 1999.
[7] HSU C, LAI Y, CHEN W, et al. Unsupervised ranking using graph structures and node attributes[C] //Proc-eedings of the Tenth ACM International Conference on Web Search and Data Mining. New York, USA: ACM, 2017: 771-779.
[8] GUO T, CAO X, CONG G, et al. Distributed algorithms on exact personalized pagerank[C] //Proceedings of the 2017 ACM International Conference on Management of Data. New York, USA: ACM, 2017: 479-494.
[9] CIPOLLA S, REDIVO-ZAGLIA M, TUDISCO F. Extrapolation methods for fixed-point multilinear PageRank computations[J]. Numerical Linear Algebra with Applications, 2020, 27(2): e2280.
[10] YANG X, WANG Q. Crowd hybrid model for pedestrian dynamic prediction in a corridor[J]. IEEE Access, 2019, 7: 95252-95261.
[11] GHOSH B, ASIF M, DAUWELS J, et al. Dynamic prediction of the incident duration using adaptive feature set[J]. IEEE Transactions on Intelligent Transportation Systems, 2018, 20(11): 4019-4031.
[12] MICHELUCCI P, DICKINSON J. The power of crowds[J]. Science, 2016, 351(6268): 32-33.
[13] NGUYEN Q, TUDISCO F, GAUTIER A, et al. An efficient multilinear optimization framework for hypergraph matching[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2016, 39(6): 1054-1075.
[14] LIN W. Distributed algorithms for fully personalized pagerank on large graphs[C] //The World Wide Web Conference. New York, USA: ACM, 2019:1084-1094.
[15] WANG W, MAZAITIS K, COHEN W. Programming with personalized pagerank: a locally groundable first-order probabilistic logic[C] //Proceedings of the 22nd ACM international conference on Information & Knowledge Management. New York, USA: ACM, 2013: 2129-2138.
[16] SAEZ T, HOGAN A. Automatically generating Wikipedia info-boxes from Wikidata[C] //Companion Proceedings of the The Web Conference 2018. Lyon, France: International World Wide Web Conferences Steering Committee. Republic and Canton of Geneva, Switzerland, 2018: 1823-1830.
[17] WANG J, WANG X, WU J. Inferring metapopulation propagation network for intra-city epidemic control and prevention[C] //Proceedings of the 24th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining. New York, USA: ACM, 2018: 830-838.
[18] 马建红,张明月,赵亚男. 面向创新设计的专利知识抽取方法[J]. 计算机应用, 2016, 36(2): 465-471. MA Jianhong, ZHANG Mingyue, ZHAO Ya'nan. Patent knowledge extraction method for innovation design[J]. Journal of Computer Applications, 2016, 36(2): 465-471.
[19] LI J. Exploring the logic and landscape of the knowledge system: multilevel structures, each multiscaled with complexity at the mesoscale[J]. Engineering, 2016, 2(3): 276-285.
[20] GLEICH D, LIM L, YU Y. Multilinear pagerank[J]. SIAM Journal on Matrix Analysis and Applications, 2015, 36(4): 1507-1541.

多维度评价

Viewed

Full text

Abstract

Cited

Shared

Discussed

基于关系转移和增强学习的时空大数据动态预测

Dynamic prediction of spatiotemporal big data based on relationship transfer and reinforcement learning

PDF (PC)

摘要/Abstract

引用本文

使用本文

参考文献

相关文章 1

多维度评价

本文评价

推荐阅读 10