[an error occurred while processing this directive]

Journal of Systems Engineering and Electronics ›› 2022, Vol. 33 ›› Issue (3): 647-664.doi: 10.23919/JSEE.2022.000060

• • 上一篇    下一篇

  

  • 收稿日期:2020-11-30 接受日期:2022-05-06 出版日期:2022-06-18 发布日期:2022-06-24

Adaptive spectral affinity propagation clustering

Lin TANG1(), Leilei SUN1,2(), Chonghui GUO1,*(), Zhen ZHANG1()   

  1. 1 Institute of Systems Engineering, Dalian University of Technology, Dalian 116024, China
    2 School of Computer Science and Engineering, Beihang University, Beijing 100191, China
  • Received:2020-11-30 Accepted:2022-05-06 Online:2022-06-18 Published:2022-06-24
  • Contact: Chonghui GUO E-mail:tanglin@dlut.edu.cn;leileisun@buaa.edu.cn;dlutguo@dlut.edu.cn;zhen.zhang@dlut.edu.cn
  • About author:|TANG Linwas born in 1980. She received her B.S. degree in computer science and technology from Liaoning Technology University in 2003, and M.S. degree in computer application from Dalian University of Technology in 2008. She is currently pursuing her Ph.D. degree on management science and engineering at Dalian University of Technology. Her research interests include text data mining, machine learning, and deep learning. E-mail: tanglin@dlut.edu.cn||SUN Leileiwas born in 1985. He received his B.S. and M.S. degrees in control theory and control engineering from Dalian University of Technology in 2009 and 2012. He received his Ph.D. degree from Institute of Systems Engineering, Dalian University of Technology, in 2017. He was a postdoctoral research fellow from 2017 to 2019 in School of Economics and Management, Tsinghua University. Currently he is an assistant professor of the State Key Laboratory of Software Development Environment and Big Data Brain Computing Lab, Beihang University. His research interests include machine learning and data mining. He has published several papers on IEEE Trans. on Data and Knowledge Engineering, Knowledge and Information Systems, and ACM SIGKDD International Conference on Knowledge Discovery and Data Mining.E-mail: leileisun@buaa.edu.cn||GUO Chonghui was born in 1973. He received his B.S. degree in Mathematics from Liaoning University in 1995, M.S. degree in Operational Research and Control Theory in 1999, and Ph.D. degree in Management Science and Engineering from Dalian University of Technology in 2002. He was a postdoctoral research fellow in the Department of Computer Science in Tsinghua University. Currently he is a professor of the Institute of Systems Engineering, Dalian University of Technology. His research concentrate on data mining and knowledge discovery. He has published over 150 peer-reviewed papers in academic journals and conferences, in addition to five text-books and two monographs. He has been the Principal Investigator on over 10 research projects from the Government and the Industry.E-mail: dlutguo@dlut.edu.cn||ZHANG Zhen was born in 1986. He received his B.S. in engineering management from China University of Petroleum (Eastern China) and Ph.D. degree in management science and engineering from the Dalian University of Technology in 2014. Currently he is an associate professor with the Institute of Systems Engineering, Dalian University of Technology. His current research interests include group decision making, computing with words and big data analysis. He is an associate editor of Kybernetes and Journal of Intelligent & Fuzzy Systems, and an editorial board member of the International Journal of Computational Intelligence Systems.E-mail: zhen.zhang@dlut.edu.cn
  • Supported by:
    This work was supported by the National Natural Science Foundation of China (71771034; 71901011; 71971039) and the Scientific and Technological Innovation Foundation of Dalian (2018J11CY009).

Abstract:

Affinity propagation (AP) is a classic clustering algorithm. To improve the classical AP algorithms, we propose a clustering algorithm namely, adaptive spectral affinity propagation (AdaSAP). In particular, we discuss why AP is not suitable for non-spherical clusters and present a unifying view of nine famous arbitrary-shaped clustering algorithms. We propose a strategy of extending AP in non-spherical clustering by constructing category similarity of objects. Leveraging the monotonicity that the clusters’ number increases with the self-similarity in AP, we propose a model selection procedure that can determine the number of clusters adaptively. For the parameters introduced by extending AP in non-spherical clustering, we provide a grid-evolving strategy to optimize them automatically. The effectiveness of AdaSAP is evaluated by experiments on both synthetic datasets and real-world clustering tasks. Experimental results validate that the superiority of AdaSAP over benchmark algorithms like the classical AP and spectral clustering algorithms.

Key words: affinity propagation (AP), Laplacian eigenmap (LE), arbitrary-shaped cluster, model selection