Xiulin Wang

Xiulin Wang

Joint-training, Ph.D candidate       Curriculum Vitae

Faculty of IT, University of Jyväskylä
Mattilanniemi 2, Building Agora B423.1,Jyväskylä, FINLAND
School of Biomedical Engineering, Dalian University of Technology
No.2, Linggong Road, Ganjingzi District, Dalian, Liaoning, China
Supervisor: Fengyu Cong , Tapani Ristaniemi
Laboratory: ASAP , SPWC
xiulin dot x dot wang at jyu dot fi
xiulin dot wang at foxmail dot com


  • 10/2017~now, Mathematical Information Technology, Faculty of Information Technology, University of Jyväskylä, Finland.
  • 09/2016~09/2017, Biomedical Engineering, School of Biomedical Engineering, Dalian University of Technology, China.
  • 07/2015~07/2016, DSP Engineer, Beijing Huiqing Techology co., China.
  • 09/2012~06/2015, M.E., Signal and Information Processing, School of Information and Communication Engineering, Dalian University of Technology, China.
  • 09/2008~06/2015, B.E., Communication Engineering, School of Electrical and Information Engineering, Shandong Unversity(Weihai), China.

Research Interest

My current research focuses on Joint analysis of multiple datasets, and its applications in the image, audio, neuroscience and biomedical engineering. My research target is to develop the new high-efficient and universal algorithms and software toolbox. Specifically:
  • 1. Tensor decomposition
  • 2. Coupled/Linked/Joint tensor decomposition
  • 3. Signal processing(e.g. brain, image, audio etc.)
  • 4. Blind Source Separation/Joint Blind Source Separation


  • Generalized Non-orthogonal Joint Diagonalization with LU Decomposition and Successive Rotations. X-F Gong, X-L Wang, Q-H Lin. IEEE Trans. Signal Process., vol. 63, no. 5, pp. 1322-1334, Mar. 2015 Download

  • A Study on Parallelization of Successive Rotation Based Joint Diagonalization. X-L Wang, X-F Gong, and Q-H Lin. Proc.DSP2014, Hong Kong, China, 2014   Download

Conference and Workshop

MATLAB Toolbox

Generalized Non-orthogonal Joint Diagonalization (GNJD) is an algorithm to simultaneously perform multiple asymmetric NJD upon multiple sets of target matrices with mutually linked loading matrices. It is a promising method to multiset data analysis problems such as joint blind source separation (J-BSS) and multiset data fusion. This software package provides source programs of GNJD to enable reproduction of the results in the published paper. We provide the routines of all the 5 experiments in the paper for our algorithms (GNJD and 2 other earlier related works). Results for other competitors, however, are not included to avoid copyright conflicts. Data and some routines for the last 2 experiments are obtained from internationally published benchmarks. We include them in the package to enable successful execution for these examples.     User Guide     Download

  • http://xiulin.wang/
  • Last updated on Sept.2018 by L