Fumin Shen
Fumin Shen(沈复民), Ph.D.
Center for Future Media
Innovation Center,  Qingshuihe Campus
    No. 2006, Xiyuan Ave, Chengdu, China
  •  fumin dot shen at gmail.com
  •  +86 61831687
cfm     uestc

I am a Professor in School of Computer Science and Engineering, University of Electronic Science and Technology of China.  I received my PhD degree from School of Computer Science, Nanjing University of Science and Technology, China in 2014. From 2010 to 2012, I visited NICTA, Australian National University and ACVT, University of Adelaide, Australia, as a joint PhD student supervised by Prof. Chunhua Shen. I received my bachelor degree in Applied Mathematics from Shandong University in 2007.

My research focuses on computer vision and machine learning, especially learning based hashing algorithms and their applications in visual retrieval and recognition problems. I have published 100+ papers in CVPR, ICCV, MM, SIGIR, TPAMI, TIP, TMM, TCSVT, etc. I have served as a regular PC member for ICCV, ACM MM, AAAI, ICMR, ICPR, MMM, journal reviewer for IEEE TIP, TNNLS, TKDE, TMM, TCYB, TCSVT, guest editor for Neurocomputing, PRL, NPL, special issue chair for ACPR'17 and special session organizer for MMM'16, ICMICS'16. I am the recipient of the Best Paper Award Honorable Mention at ACM SIGIR 2016 & ACM SIGIR 2017 and the World's FIRST 10K Best Paper Award - Platinum Award at IEEE ICME 2017.

  Students Vacancies(招生)

I am always looking for highly motivated students or interns. Please do not hesitate to send me your CV, if you are interested in computer vision, multimedia, machine learning research.



Professor University of Electronic Science and Technology of China, China 2017.07 ~ present
Associate Professor University of Electronic Science and Technology of China, China 2016.08 ~ 2017.06
Lecturer University of Electronic Science and Technology of China, China 2015.03 ~ 2016.07
Joint Ph.D. ACVT, University of Adelaide, Australia 2011.06 ~ 2012.11
Joint Ph.D. NICTA, Australian National University, Australia 2010.11 ~ 2011.05
Ph.D. Nanjing University of Science and Technology, China 2007.09 ~ 2014.04
Bachelor Shandong University, China 2003.09 ~ 2007.07


Highlighted work More projects...

Recently, learning based hashing techniques have attracted broad research interests because they can support efficient storage and retrieval for high-dimensional data such as images, videos, documents, etc. However, a major difficulty of learning to hash lies in handling the discrete constraints imposed on the pursued hash codes, which typically makes hash optimizations very challenging (NP hard in general). The widely adopted relaxation scheme simplifies the optimization, however, tends to produce less effective codes.

In this project, we investigate how to directly learn binary codes via discrete optimization, namely discrete hashing. Discrete hashing has been shown to significantly boost the performance of hashing algorithms (e.g., Supervised Discrete Hashing) in large-scale similarity search.


Most binary coding and hashing algorithms were designed to deal with Approximate Nearest Neighbor (ANN) search. Studies have rarely been dedicated to Maximum Inner Product Search (MIPS), which actually plays a critical role in various vision and learning applications.

In this project, we investigate learning binary codes to exclusively handle the MIPS problem. The core problem is, the similaritis between two sets of data should be revealed by the inner products of their corresponding binary codes. We study this problem by proposing the Asymmetric Inner-product Bianry Coding algorithm for large-scale MIPS search, and the Discrete Collaborative Filtering algorithm for recommendation systems.

In the recent SIGIR paper, we study asymmetric hashing on the large-scale linear classification problem: classify examples by a linear transformation matrix from a large number of categories. This is a typical inner-product involving problem. We propose to respresent both the samples and classifiers using binary hash codese, which are simultaneously learned from the training data. Classifying an example thereby reduces to retrieving its nearest class codes in the Hamming spac, i.e., computing the Hamming distance between the binary codes of the example and classifiers and selecting the class with minimal Hamming distance.


Recent publications More papers...