Runner to Tensor Learning. | Founder of White Paper Tao (WPT).

Name: Andong Wang (王安东)
Email: w.a.d@outlook.com antonw@qq.com
Experience: 2008 ~ 2012, NJUST| Software Engineering, Bacholar 2012 ~ , NJUST| Pattern Recognition, PHD

Hi, my name is Andong Wang, a 7th-year Phd Candidate majored in Pattern Recognition Theory and Application in NJUST. I was born in Linqu county(临朐县), Shandong Province .


I am supervised by Professor Zhong Jin. I am interested in the intersection of low-rank tensor decomposition and high-dimentional statistics.

Research Experience

2012.9 ~ 2014.12 Negative Obstacle Detection for Unmanned Vehicle

Before studying the tensor learning problem, I had about two years' research experience in environment sensing for unmanned ground vehicle. Through designing negative obstacle detection algorithms and implementing them in on-board embedded systems, I obtained a strong practical ability on programming.

2015.1 ~ Low-rank Tensor Decomposition

Currently, I am a core member in a general project supported by the Natural Science Foundation of China, titled "Research on tensor learning theory and algorithms based on low tubal rank structure". My previous work mainly focuses on low rank tensor recovery, like tensor completion and tensor robust principle component analysis, which aims to recover a low rank tensor from its partial and/or noisy observations. In my research papers, usually a novel model is proposed to estimate the underlying signal tensor, along with an optimization algorithm to efficiently solve it. Then, the statistical performance of the proposed model is analyzed by establishing deterministic/non-asymptotic upper bounds on the estimation error. The proposed models find many applications in real-world multi-way data recovery tasks, such as image/video inpainting and de-noising.

Skills (with score in percentage)

Basic Tensor Algebra 80%
High-dimentional Statistics 75%
Convex Optimization 75%
Coding (C++, Matlab, Java, Python) 85%
Experimental Skills 75%
Stickability 85%

Published Papers

[8] Andong Wang, Zhihui Lai, Zhong Jin. Noisy low-tubal-rank tensor completion. Neurocomputing. DOI: https://doi.org/10.1016/j.neucom.2018.11.012. [Pdf link]
[7] Andong Wang, Dongxu Wei, Bo Wang, Zhong Jin. Noisy low-tubal-rank tensor completion through iterative singular tube thresholding. IEEE Access, vol. 6, pp. 35112–35128, 2018. [Pdf link]
[6] Dongxu Wei, Andong Wang, Bo Wang, Xiaoqin Feng. Tensor completion using spectral (k,p)-support norm. IEEE Access, vol. 6, pp. 11559–11572, 2018. [Pdf link]
[5] Dongxu Wei, Andong Wang, Xiaoqin Feng, Boyu Wang, Bo Wang. Tensor completion based on triple tubal nuclear norm. Algorithms, vol. 11(7), 2018. [Pdf link]
[4] Andong Wang, Zhong Jin. Near-optimal noisy low-tubal-rank tensor completion via singular tube thresholding. ICDM Workshops, 2017, pp. 553–560. [Pdf link]
[3] Andong Wang, Zhong Jin. Coherent low-tubal-rank tensor completion. ACPR, 2017, pp. 518-523. [Pdf link]
[2] Xiangrui Li, Andong Wang, Jianfeng Lu, Zhenmin Tang. Statistical performance of convex low-rank and sparse tensor recovery. ACPR, 2017, pp. 524-529.
[1] Jiayin Liu, Zhenmin Tang, Andong Wang, Chaoxia Shi. Negative obstacle detection in unstructured environment based on multiple lidars and compositional features. ROBOT, vol. 39(5), pp. 638-651, 2017. (in Chinese) [Pdf link]

Dratfs (with theory ready)

[8] Andong Wang, Xulin Song, Xiyin Wu, Zhihui Lai, Zhong Jin. Generalized Dantzig selector for low-tubal-rank tensor recovery.(submitted)
[7] Andong Wang, Xulin Song, Xiyin Wu, Zhihui Lai, Zhong Jin. Robust low-tubal-rank tensor completion. (submitted)
[6] Andong Wang, Xulin Song, Xiyin Wu, Zhihui Lai, Zhong Jin. Latent Schatten TT norm for tensor completion. (submitted)
[5] Andong Wang, Zhong Jin. Orientation invariant tubal nuclear norms applied to robust tensor decomposition. preprint. DOI: 10.13140/RG.2.2.32600.55041. [Pdf link]
[4] Xiangrui Li, Andong Wang, Jianfeng Lu, Zhenmin Tang. Statistical performance of convex low-rank and sparse tensor recovery. Pattern Recognition (major revision).
[3] Variational Bayesian N-way Partial Least Squares.
[2] Low-rank Tensor Recovery under Heavy-tailed Noise.
[1] Low-rank Tensor Recovery via Overlapped Group Sparse Tucker Core.

Five-year Direction

Low-rank Tensor Network & High-dimentional Statistics

站长统计