Centre for Quantum Computation and Intelligent Systems (QCIS)
Faculty of Engineering and Information Technology (FEIT)
University of Technology, Sydney (UTS)
Level 3, Building 10
235 Jones Street, Broadway
NSW 2007, Australia

Tianyi Zhou is with Centre for Quantum Computation & Intelligent Systems (QCIS)Faculty of Engineering & ITUniversity of Technology Sydney (UTS) in Australia, supervised by Prof. Dacheng Tao. He received his B.S. degree in automatic control from Beijing Institute of Technology (BIT) in 2008. His research interests are in machine learning, compressed sensing, matrix completion and factorization, dimension reduction and fast optimization algorithms.

Previously, he was a research assistant in School of Computer Eigneering at Nanyang Technological University (NTU) in Singapore, supervised by Prof. Dacheng Tao, between Jan 2009-Jan 2011. Before that, he was a research assistant at Institute of Automation, Chinese Academy of Science (CASIA) in Prof. Tieniu Tan’s group.

His current topics include compressed sensing with quantization, sparse dimension reduction, model selection, manifold learning, multi-label learning, covariance selection, low-rank and sparse matrix decomposition, matrix completion, Nesterov's gradient method and its extensions, fast gradient clustering, low-rank learning, graphical model selection/learning, dual decomposition and non-parametric Bayesian.

For specific information, please refer to his publications, CV, awards, Google Scholar and research blog.

Education:       BS:     Automatic Control, Beijing Institute of Technology, Beijing, P. R. China, 2004-2008.

Experience:     RA:     Machine Learning, University of Technology, Sydney, Australia, 2011-Present.
                            RA:     Machine Learning, Nanyang Technological University, Singapore, 2009-2011.
                            RA    Institute of Automation, Chinese Academy of Science, Beijing, P. R. China, 2008-2009.

Our IJCAI 2013 paper "Shifted Subspace Tracking on Sparse Outliers for Motion Segmentation" PDF is online. to find out How to extend GoDec in nonlinear low-rank case? How to develop a simple matrix factorization method for complex computer vision tasks such as detection, tracking and motion segmentation. I will give an oral presentation on IJCAI 2013 Beijing, if you are interested, please join me at BICC Room 305ABC between 10:15-10:35, August 9th 2013 (Friday Morning).  July 05,2013

Our ISIT 2013 paper "k-bit Hamming Compressed Sensing" PDF is online. to find out How to recover d-level quantization of signal from k-bit quantization of linear measurements via fast hypothesis testing in parallel. I will give an oral presentation on ISIT 2013 Istanbul Turkey, if you are interested, please join me at ICEC Room Topcapi A between 17:50-18:10, July 8th 2013 (Monday Afternoon).   July 05,2013

Our DMKD paper about Manifold Elastic Net (MEN) has been selected as Top 5 Editor's Choice paper for Free Reading, May 10, 2013.  

k-bit Hamming Compressed Sensing, a nontrivial extension of 1-bit Hamming Compressed Sensing that recovers d-bit signal from its k-bit quantization directly by hypothesis testing, is accepted by ISIT 2013, Apr 12, 2013

Shifted Subspaces Tracking on Sparse Outlier for Motion Segmentation, which reduces motion detection, tracking and segmentation to one efficient matrix factorization model (extension of GoDec), is accepted by IJCAI 2013, Apr 02, 2013

Greedy Bilateral Sketch, Completion and Smoothing for large-scale matrix completion, robust PCA and low-rank approximation is accepted by AISTATS 2013, here comes brief introductionPDF can be found here.  Feb 11, 2013

Double Shrinking for Sparse Dimension Reduction is accepted by IEEE Trans on Image Processing.     May 13, 2012

Bilateral Random Projections and 1-bit Hamming Compressed Sensing are accepted as presentations on IEEE International Symposium on Information Theory (ISIT) 2012.     Apr 18, 2012

Bilateral Random Projections preprint is now available on arxiv. PDF is here.     Dec 22, 2011

Multi-label Subspace Ensemble (Related to GoDec) is accepted by International Conference on Artificial Intelligence and Statistics (AISTATS) 2012. PDF can be found here.      Dec 12, 2011

Compressed Labeling on Distilled Labelsets for Multi-label Learning (Related to HCS) is accepted by Machine Learning Journal (Springer). PDF can be downloaded here.       Dec 11, 2011     

 Semi-Soft GoDec MatLab code is released. This variant of GoDec algorithm is much faster and need not to artificially select a good parameter k.    Nov 14, 2011

Hamming Compressed Sensing webpage is open, you can download the paper and find experimental results on it.     Oct 04, 2011

GoDec website is open, you can download the code and find the relevant videos from it.     Aug 29, 2011