Ha Quang Minh

Functional Analytic Learning Unit
RIKEN Center for Advanced Intelligence Project, Tokyo

I am currently a Unit Leader (equivalent to Associate Professor) at the RIKEN Center for Advanced Intelligence Project (RIKEN-AIP) in central Tokyo, Japan, where I lead the Functional Analytic Learning Unit. Prior to joining the RIKEN-AIP, I was a researcher at the Pattern Analysis and Computer Vision (PAVIS) group, at the Istituto Italiano di Tecnologia (IIT) – Italian Institute of Technology, in Genova (Genoa), Italy. I received my PhD in Mathematics from Brown University, Providence, RI, USA, under the supervision of Steve Smale, and my dissertation was on Reproducing Kernel Hilbert Spaces (RKHS). I used to visit France during the jours feries.

I am interested in both the mathematical foundations and algorithmic developments in machine learning, AI, computer vision, and image and signal processing, and problems in applied and computational functional analysis, and applied and computational differential geometry.

My current research focuses on the following two principal directions, which are closely related :

  1. Functional analytic methods in machine learning, including in particular methods from matrix and operator theory, and the theory of vector-valued Reproducing Kernel Hilbert Spaces (RKHS)
  2. Geometrical methods in machine learning, including in particular methods from Riemannian geometry and related areas. My current focus is on the geometry of RKHS covariance operators and their applications.

About Me

Education

Previous Academic Affiliations


Selected Recent Publications

RKHS in the framework of Riemannian geometry and applications

The following papers study the infinite-dimensional Hilbert manifold of positive definite operators on a Hilbert space, with a particular focus on RKHS covariance operators.

(Refereed conference paper) Hà Quang Minh. Infinite-Dimensional Log-Determinant Divergences III: Log-Euclidean and Log-Hilbert–Schmidt Divergences. In: Information Geometry and its Applications IV. pp. 209-243. Springer (2018)

(Tutorial) From Covariance Matrices to Covariance Operators: Data Representation from Finite to Infinite-Dimensional Settings, at ICCV 2017, Venice, October 2017.

(Book) Hà Quang Minh and V. Murino. Covariances in Computer Vision and Machine Learning, Morgan & Claypool Publishers, October 2017.

(Journal article) D. Felice, M. Hà Quang, S. Mancini. The volume of Gaussian states by information geometry. Journal of Mathematical Physics, 58(1): 012201, 2017.

(Refereed conference paper) Hà Quang Minh. Log-Determinant divergences between positive definite Hilbert-Schmidt operators, Geometric Science of Information, November, 2017.

(Journal article) Hà Quang Minh. Infinite-dimensional Log-Determinant divergences between positive definite trace class operators, Linear Algebra and Its Applications, volume 528, pages 331-383, September, 2017.

(Book) Hà Quang Minh and V. Murino (editors). Algorithmic Advances in Riemannian Geometry and Applications, Springer series in Advances in Computer Vision and Pattern Recognition, 2016.

(Book chapter) Hà Quang Minh and V. Murino. From Covariance Matrices to Covariance Operators: Data Representation from Finite to Infinite-Dimensional Settings. In Algorithmic Advances in Riemannian Geometry and Applications, Springer, 2016.

(Refereed conference paper) Hà Quang Minh, M. San Biagio, L. Bazzani, V. Murino. Approximate Log-Hilbert-Schmidt distances between covariance operators for image classification. IEEE Conference on Computer Vision and Pattern Recognition (CVPR 2016), Las Vegas, USA, June 2016.

(Refereed conference paper) Hà Quang Minh. Affine-invariant Riemannian distance between infinite-dimensional covariance operators. Geometric Science of Information (GSI 2015), Paris, France, October 2015.

(Refereed conference paper) Hà Quang Minh, Marco San Biagio and Vittorio Murino. Log-Hilbert-Schmidt metric between positive definite operators on Hilbert spaces. Advances in Neural Information Processing Systems (NIPS 2014), Montreal (Canada), December 2014. Supplementary Material.

Vector-valued RKHS and applications

The following paper introduces a general learning formulation in vector-valued RKHS that encompasses many learning algorithms in the literature in a single framework. Examples include vector-valued least square regression and classification, multi-class SVM (using the Simplex Coding), and multi-modality (multi-view/multi-feature) learning, in both the supervised and semi-supervised settings.

(Journal article) Hà Quang Minh, L. Bazzani, V. Murino. A Unifying Framework in Vector-valued Reproducing Kernel Hilbert Spaces for Manifold Regularization and Co-Regularized Multi-view Learning. Journal of Machine Learning Research, 17(25):1-72, 2016.


Publications

Recent and Upcoming Publications

2016

Hà Quang Minh, M. San Biagio, L. Bazzani, V. Murino. Approximate Log-Hilbert-Schmidt distances between covariance operators for image classification. IEEE Conference on Computer Vision and Pattern Recognition (CVPR 2016), Las Vegas, USA, June 2016.

Hà Quang Minh, L. Bazzani, V. Murino. A Unifying Framework in Vector-valued Reproducing Kernel Hilbert Spaces for Manifold Regularization and Co-Regularized Multi-view Learning. Journal of Machine Learning Research, 17(25):1-72, 2016.

Main paper (72 pages): 14-036

2015

Hà Quang Minh. Affine-invariant Riemannian distance between infinite-dimensional covariance operators. Geometric Science of Information (GSI 2015), Paris, France, October 2015.

L. Dodero, Hà Quang Minh, M. San Biagio, V. Murino and D. Sona
Kernel-based Classification For Brain Connectivity Graphs On The Riemannian Manifold Of Positive Definite Matrices. International Symposium on Biomedical Imaging (ISBI 2015), New York, USA, April 2015.

2014

Hà Quang Minh, Marco San Biagio and Vittorio Murino. Log-Hilbert-Schmidt metric between positive definite operators on Hilbert spaces. Advances in Neural Information Processing Systems (NIPS 2014), Montreal (Canada), December 2014.

Complete List of Publications

PhD Dissertation

Reproducing Kernel Hilbert Spaces in Learning Theory, advisor: Steve Smale, Department of Mathematics, Brown University, Providence, RI, USA, May 2006.

PDF: thesisMinh2006

Journal Articles

Hà Quang Minh, L. Bazzani, V. Murino. A Unifying Framework in Vector-valued Reproducing Kernel Hilbert Spaces for Manifold Regularization and Co-Regularized Multi-view Learning. Journal of Machine Learning Research, 17(25):1-72, 2016.
Main paper (72 pages): 14-036

Hà Quang Minh and Laurenz Wiskott. Multivariate slow feature analysis and decorrelation filtering for blind source separation. IEEE Transactions on Image Processing, volume 22, issue 7, pages 2737-2750, July 2013.
Main paper: IEEE_TIP_SFA_MinhWiskott_2013

Supplementary material: Ha Quang Minh. The regularized least squares algorithm and the problem of learning halfspaces, Information Processing Letters, volume 111, issue 8, pages 395-401, March 2011.

Paper: https://www.tables-de-multiplication.com

Gianluigi Pillonetto, Minh Ha Quang and Alessandro Chiuso. A New Kernel-based Approach for Nonlinear System Identification, IEEE Transactions on Automatic Control, volume 56, issue 12, pages 2825-2840, December 2011.

Ha Quang Minh. Some Properties of Gaussian Reproducing Kernel Hilbert Spaces and Their Implications for Function Approximation and Learning Theory, Constructive Approximation, volume 32, number 2, pages 307-338, 2010.

Paper: gaussianpaper_minh_official_2010

Minh Ha Quang, Sung Ha Kang, and Triet Le, Image and video colorization using vector-valued reproducing kernel Hilbert spaces, Journal of Mathematical Imaging and Vision, volume 37, number 1, pages 49-65, 2010.

Paper: colorpaper_Minh2010_official

Refereed Papers in International Conference Proceedings

2016

Hà Quang Minh, M. San Biagio, L. Bazzani, V. Murino. Approximate Log-Hilbert-Schmidt distances between covariance operators for image classification. IEEE Conference on Computer Vision and Pattern Recognition (CVPR 2016), Las Vegas, USA, June 2016, to appear.

2015

Hà Quang Minh. Affine-invariant Riemannian distance between infinite-dimensional covariance operators. Geometric Science of Information (GSI 2015), Paris, France, October 2015.

L. Dodero, Hà Quang Minh, M. San Biagio, V. Murino and D. Sona
Kernel-based Classification For Brain Connectivity Graphs On The Riemannian Manifold Of Positive Definite Matrices. International Symposium on Biomedical Imaging (ISBI 2015), New York, USA, April 2015.

2014

Hà Quang Minh, Marco San Biagio and Vittorio Murino. Log-Hilbert-Schmidt metric between positive definite operators on Hilbert spaces. Advances in Neural Information Processing Systems (NIPS 2014), Montreal (Canada), December 2014.

Main paper: LogHilbert_NIPS2014_final

Supplementary material:LogHilbert_NIPS2014_Supplementary_final

2013

V.Sindhwani, H.Q. Minh and A.C. Lozano. Scalable Matrix-valued Kernel Learning for High-dimensional Nonlinear Multivariate Regression and Granger Causality. Proceedings of the 29th Conference on Uncertainty in Artificial Intelligence (UAI 2013), July 2013, Bellevue, Washington, USA (Microsoft Best Paper Award).

Hà Quang Minh, Loris Bazzani, and Vittorio Murino. A unifying framework for vector-valued manifold regularization and multi-view learning. Proceedings of the 30th International Conference on Machine Learning (ICML 2013), June 2013, Atlanta, Georgia, USA.

Main paper: ICML_2013_MinhBazzani

Supplementary material: ICML_2013_MinhBazzani_Supp

D. Figueira, L. Bazzani, H.Q. Minh, M. Cristani, A. Bernardino, and V. Murino. Semi-supervised multi-feature learning for person re-identification. Proceedings of the 10th IEEE International Conference on Advanced Video and Signal-based Surveillance (AVSS 2013), Krakow, Poland, August 2013.

G. Roffo, M. Cristani, L. Bazzani, H.Q. Minh, V. Murino. Trusting Skype: Learning the Way People Chat for Fast User Recognition and Verification. IEEE Workshop in Decoding Subtle Cues from Social Interactions, in conjunction with ICCV 2013, Sydney, Australia, December 2013.

2012

Hà Quang Minh, Marco Cristani, Alessandro Perina and Vittorio Murino. A regularized spectral algorithm for Hidden Markov Models with applications in computer vision. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR 2012), June 2012, Providence, RI, USA.

2011

Hà Quang Minh and Laurenz Wiskott. Slow feature analysis and decorrelation filtering for separating correlated sources. Proceedings of the 13th IEEE International Conference on Computer Vision (ICCV 2011), November 2011, Barcelona, Spain.

Paper: iccv_paper_511_final_minh

Supplementary material: iccv_paper_511_supplementary_material_final

Hà Quang Minh and Vikas Sindhwani. Vector-valued Manifold Regularization. Proceedings of the 28th International Conference on Machine Learning (ICML 2011), June 2011, Bellevue, Washinton, USA.

Paper: Minh_Sindhwani_ICML2011_VVMR

2009

Minh Ha Quang, Gianluigi Pillonetto and Alessandro Chiuso. Nonlinear system identification via Gaussian regression and mixtures of kernels, Proceedings of the 15th IFAC Symposium on System Identification (SYSID 2009), Saint-Malo, France, July 2009.

Minh Ha Quang, Sung Ha Kang, and Triet Le. Reproducing kernels and colorization, Proceedings of the 8th International Conference on Sampling Theory and Applications (SAMPTA 09), Luminy, France, May 2009.

2006

Ha Quang Minh, Partha Niyogi and Yuan Yao. Mercer’s Theorem, Feature Maps, and Smoothing, Proceedings of the 19th Annual Conference on Learning Theory (COLT 2006), Springer Lecture Notes in Computer Science volume 4005, pages 154-168.

Paper: Minh_MercerTheorem_COLT2006

2004

Ha Quang Minh and Thomas Hofmann. Learning over Compact Metric Spaces, Proceedings of the 17th Annual Conference on Learning Theory (COLT 2004),Springer Lecture Notes in Artificial Intelligence volume 3120, pages 239-254.

Paper: Minh_LearningCompactSpaces_COLT2004

Awards

UAI 2013 Microsoft Best Paper Award, for joint work with Vikas Sindhwani and Aurelie Lozano.

IBM Pat Goldberg Memorial Best Paper Award in Computer Science, Electrical Engineering and Mathematics 2013, for joint work with Vikas Sindhwani and Aurelie Lozano.

Open positions

Postdoctoral positions at the RIKEN Center for Advanced Intelligence Project

The Functional Analytic Learning Unit at the RIKEN Center for Advanced Intelligence Project (RIKEN-AIP) is looking for Postdoctoral Researchers to join our team. The unit focuses on functional analytic and geometrical methods both in fundamental machine learning research and their applications.

Current machine learning projects include mathematical foundations of machine learning, geometrical data analysis, functional data analysis, covariance matrices and covariance operators, matrix-valued and operator-valued kernels, approximation methods for big data, manifold learning, multi-modality learning, and multi-task learning.

Current targeted application domains include, but are not limited to, computer vision, image and signal processing, and brain imaging.

How to apply

Please follow instructions given at

http://www.riken.jp/en/careers/researchers/20171205_3/

For further information, please email minh dot haquang at riken dot jp

and/or visit

https://aip.riken.jp/labs/generic_tech/funct_anl_learn/

About the Center for Advanced Intelligence Projects (AIP)

The Center for Advanced Intelligence Project has been launched with subsidy from Ministry of Education, Culture, Sports, Science and Technology-Japan for “Advanced Integrated Intelligence Platform Project (AIP) -Artificial Intelligence/ Big Data/ Internet of Things/ Cybersecurity-”.

The center is home to a large number of experts in machine learning and related fields (mathematics, optimization, statistics, life science, image processing, natural language processing, etc.). The center’s headquarters and many of its teams are located in the Tokyo area, but many other teams are scattered all over Japan (Kyoto, Nagoya, Osaka, Sendai, Kyushu etc.).

For more information on the AIP, please visit https://aip.riken.jp/

About the RIKEN

RIKEN is the largest comprehensive research institution in Japan, encompassing a network of world-class research centers and institutes across Japan.

For more information on the RIKEN, please visit http://www.riken.jp/en/