Namgil Lee

                  Kangwon National University
                  Chuncheon, Gangwon, 14341, Republic of Korea

E-mail: namgil.lee(at) / namgil.lee(at)

Tel: +82-33-250-8433
Fax: +82-33-259-5663

Research Interests

  • High-dimensional Statistical Modeling
Understanding and modeling interactions between objects / features / factors can provide more insight into the nature underlying observations and data, than simply conducting analyses with individual features. For such multivariate or high-dimensional data analysis, my research interests include
    • Graphical models
    • Multivariate time series models, e.g., vector autoregressive (VAR) models
  • Tensor decomposition for big data signal processing
The so-called tensor methods are newly arising fields of research in machine learning, signal processing, and numerical mathematics, due to their ability to process large-scale optimization problems via distributed representations. Strong mathematical theories are also growing fast to support why and how they works well. Tensor methods are also quite promising in the era of big data and AI, because they provide good mathematical senses of why recent AI algorithms are working very well.  My interests include
    • Tensor networks, e.g., tensor train (TT), for applications in large scale optimization, high-dimensional probability estimation, Monte Carlo simulation
    • Multiway component analysis (MCA)
    • Tensor completion, i.e., missing data estimation
  • Other research interests: 
    • Machine Learning: Theories and Algorithms
    • Neuroscience (FMRI, EEG) data analysis of functional brain connectivity
    • Time Series Analysis


  • New Affiliation Recently I joined the Department of Information Statistics, Kangwon National University, Chuncheon, Republic of Korea. Chuncheon is a central city of Ganwon province, and it is located near Seoul. It ...
    2017. 10. 23. 오전 7:48에 Namgil Lee님이 게시
  • "Excellent Paper Award" at the ICONIP 2016 Our paper "Nonnegative Tensor Train Decompositions for Multi-Domain Feature Extraction and Clustering" received the "Excellent Paper Award"at the ICONIP 2016, 16-21 October 2016, Kyoto, Japan. This is ...
    2016. 10. 24. 오전 6:08에 Namgil Lee님이 게시
  • ICONIP2016, Kyoto, Japan, 16-21 Oct 2016 I will participate in the conference: The 23rd International Conference on Neural Information Processing (ICONIP2016). Our title of the presentation is "Nonnegative Tensor Train Decompositions for Multi-Domain Feature Extraction ...
    2016. 9. 5. 오후 11:24에 Namgil Lee님이 게시
7개의 게시물 중 1 - 3 더보기 »


  • Singular Value Decomposition Using Tensor Train (N.Lee & A.Cichocki, 2015) ALS and MALS algorithms for EVD/SVD in MATLAB - v201502REFERENCE:Namgil Lee, Andrzej Cichocki. Estimating a Few Extreme Singular Values and Vectors for Large-Scale Matrices in Tensor Train ...
    2015. 8. 25. 오전 7:23에 Namgil Lee님이 게시
  • Conversion of Categorical Variables (N. Lee & J.-M. Kim, 2010) Matlab software for conversion of mixed categorical and numerical variables into numerical variables for binary classificationREFERENCE:Namgil Lee and Jong-Min Kim, Conversion of categorical variables into numerical variables ...
    2016. 3. 17. 오후 7:50에 Namgil Lee님이 게시
2개의 게시물 중 1 - 2 더보기 »