The goal of the workshop is to share the potential of subspace methods with researchers working on various problems in computer vision and pattern recognition, and to encourage interactions which could lead to further developments of subspace methods. Both the fundamental theories of subspace methods and their applications in computer vision will be discussed at the workshop.


Subspace methods have been used as a practical methodology in a large variety of real applications. Also they have been studied intensively, in particular, in the field of character recognition, contributing to a number of commercial optical character recognition systems. During the last four decades, the area has become one of the most successful underpinnings of diverse applications such as classification, recognition, pose estimation, motion estimation, etc. At the same time, there are many new and evolving research topics: nonlinear methods including kernel methods, manifold learning, subspace update and tracking.

We expect that the workshop could accelerate the stream, by providing the place to share the potential of subspace methods with researchers working on various problems in computer vision and pattern recognition, and encouraging interactions which could lead to further developments of subspace methods.


The topics of interest include, but are not limited to, the following:

  • Theory on Subspace methods:
    • PCA, LDA, CCA, FA, ICA, etc (linear, nonlinear/kernelized), multivariate analysis
    • Tensor and multilinear representations
    • CLAFIC, mutual subspace methods, eigenspace methods
    • Nonlinear subspace methods, including kernel methods
    • manifold learning
    • similarity measures with subspaces
    • Iijima equation, degenerated Gaussians, geometry of subspaces, etc.
  • Subspaces in various problems:
    • factorization methods
    • 3D geometry
    • local features
    • photometry and illumination constraints
    • analytic manifolds, etc.
  • Applications:
    • Object recognition
    • face recognition
    • gesture recognition
    • character recognition
    • motion analysis
    • scene analysis
    • robot vision
    • biometrics
    • anomaly detection
    • data visualization
    • other novel applications.


  • 08:00 - 09:55 Registration and Posters

  • 09:55 - 10:00 Opening

  • 10:00 - 11:00 Dimensionality reduction (3 orals, 20 minutes each)
    • High Dimensional Correspondences from Low Dimensional Manifolds - An Empirical Comparison of Graph-based Dimensionality Reduction Algorithms, Ribana Roscher, Falko Schindler, Wolfgang Forstner
    • Multi-Label Classification for Image Annotation via Sparse Similarity Voting, Tomoya Sakai, Hayato Itoh, Atsushi Imiya
    • Centered Subset Kernel PCA for Denoising, Yoshikazu Washizawa, Masayuki Tanaka

  • 11:00 - 11:10 Morning break and Posters

  • 11:10 - 12:10 Foundations of subspace method (3 orals, 20 minutes each)
    • On the Behavior of Kernel Mutual Subspace Method, Hitoshi Sakano, Osamu Yamaguchi, Tomokazu Kawahara, Seiji Hotta
    • Compound Mutual Subspace Method for 3D object recognition: A theoretical extension of Mutual Subspace Method, Naoki Akihiro, Kazuhiro Fukui
    • Dynamic Subspace Update with Incremental Nystrom Approximation, Hongyu Li, Lin Zhang

  • 12:10 - 14:10 Poster Session (Oral-Poster-Hybrid) and Lunch
  • All oral papers are presented at the poster session

  • 14:10 - 15:30 Feature Extraction / Recognition (4 orals, 20 minutes each)
    • Background Modeling via Incremental Maximum Margin Criterion, Thierry Bouwman, Cristina Marghes
    • Trace norm regularization and application to tensor based feature extraction, Yoshikazu Washizawa
    • Fast and Robust Face Recognition for Incremental Data, I Gede Pasek Suta Wijaya, Keiichi Uchimura, Gou Koutaki
    • Extracting Scene-dependent Discriminant Features for Enhancing Face Recognition under Severe Conditions, Rui Ishiyama, Nobuyuki Yasukawa

  • 15:30 - 16:00 Coffee Break and Posters

  • 16:00 - 16:50 Invited Talk
    • A Brief History of the Subspace Methods, Hitoshi Sakan

  • 16:50 - 17:00 Closing

See the presentation instruction of ACCV2010.


Online Submission

HERE is the online submission site.

Instructions and Authors' kit

Paper format should be the same as ACCV2010 (LNCS style). Page limit is 10 pages (shorter than ACCV papers) with no extra pages. Please refer the ACCV2010 author guidelines and author kit available at ACCV2010 website.

Reviewing is double-blind: remove names and affiliations of authors from the submitted paper as indicated by the ACCV2010 author guidelines.

Dual Submissions

By submitting a manuscript to Subspace 2010, the authors assert that it has not been previously published in substantially similar form. Furthermore, no paper which contains significant overlap with the contributions of this paper either has been or will be submitted during the Subspace 2010 review period to either a journal or a conference.

However, you can submit a paper to both ACCV and subspace as described below:
ACCV dual submission policy: Author(s) may submit the same manuscript also to one (and not more than one) of the workshops accompanying ACCV 2010, with informing with their submission about the preferred option (i.e., in case that accepted both to ACCV and the accompanying workshop), and also with informing in their workshop submission about the dual submission to the main conference.

Program committee

  • Organizers
    • David Suter (The University of Adelaide, Australia)
    • Kazuhiro Fukui (University of Tsukuba, Japan)
    • Toru Tamaki (Hiroshima University, Japan)
  • Program committee
    • Toshiyuki Amano (NAIST, Japan)
    • Horst Bischof (TU Graz, Austria)
    • Seiji Hotta (Tokyo University of Agriculture and Technology, Japan)
    • Masakazu Iwamura (Osaka Prefecture University, Japan)
    • Tae-Kyun Kim (University of Cambridge, UK)
    • Xi Li (Xi'an Jiaotong University, China)
    • Yi Ma (University of Illinois at Urbana Champaign, USA)
    • Atsuto Maki (Toshiba Cambridge Research Lab., UK)
    • Shinichiro Omachi (Tohoku University, Japan)
    • Bisser Raytchev (Hiroshima University, Japan)
    • Peter Roth (TU Graz, Austria)
    • Hitoshi Sakano (NTT CS Laboratories, Japan)
    • Atsushi Sato (NEC, Japan)
    • Yoichi Sato (The University of Tokyo, Japan)
    • Shin'ichi Satoh (National Institute of Informatics, Japan)
    • Terence Sim (National University of Singapore, Singapore)
    • Bjorn Stenger (Toshiba Cambridge Research Lab., UK)
    • Qi Tian (University of Texas at San Antonio, USA)
    • Fernando De la Torre (Carnegie Mellon University, USA)
    • Seiichi Uchida (Kyushu University, Japan)
    • Osamu Yamaguchi (Toshiba, Japan)
    • Jakob Verbeek (INRIA Rhone-Alpes, Grenoble, France)
    • Jing-Hao Xue (University College London, UK)