- Multivariate Statistics and Data Analysis: principal components, factor analysis, multidimensional scaling, clustering, discrimination and
classification; Procrustes analysis; three-mode methods (common principal components, INDSCAL, GIPSCAL, etc);
Statistical Learning and Data Mining: mainly unsupervised learning methods as penalized versions of principal components and discriminant analysis;
independent components analysis; approximate but very fast versions of the standard multivatiate techniques applicable for large data sets; applications to
shape, image, handwritten character recognition; multivariate analysis of microarray data;
Scientific Computing: matrix computing; computational constrained optimization on matrix manifolds, gradient projection and continuous-time optimization
methods; gradient dynamical systems on matrix manifolds; numerical integration of initial value problems for ordinary differential equations.
I organised (with Kohei Adachi, Osaka University) and chaired the Special Session of invited talks on Sparse dimension reduction at the 5th International
Conference of the ERCIM Working Group on Computing and Statistics, held 1-3 December 2012, Oviedo, Spain.
Details of the four talks are below: