It is well known that one difficulty facing today statisticians is the increasingly complex structures of datasets. Sophisticated data collection facilities often produce data which are a set of functions, represented in the form of curves, images or shapes. Recently functional data analysis, which deals with the analysis of such curves, images and shapes, has received considerable attention in the statistical literature. We gather four promising statisticians from various research areas and three different countries (Israel, Canada, U.S.A.) to discuss newly merged functional data. In this session, a good mixture of practical and theoretical talks showcases novel findings and cutting-edge achievements in functional and high-dimensional data analysis. In view of the variety of topics covered by the talks with emphasis on applications to fields including neuroscience, public health and pharmaceutical statistics, the session will attract a wide audience consisting of mathematical statisticians, epidemiologists, data scientists and other applied researchers from industry and government. 

In this session, we gather four promising statisticians from various research areas and three different countries (Israel, Canada, U.S.A.) to discuss newly merged functional and high-dimensional data. Specifically, the four speakers will present functional data classification based on the deep learning method and reproducing kernel Hilbert space framework. They will show that the proposed classifiers can achieve upper bound of the minimax optimality under mild assumptions. Furthermore, continuous-time multivariate analysis and penalized empirical likelihood with the sparse cox regression model will also be discussed for complex high-dimensional data analysis.
 

Organiser: Dr Guanqun Cao  

Chair: Dr Honglang Wang 

Speaker: Chi-Kuang Yeh 

Speaker: Philip Reiss 

Speaker: Dr Guanqun Cao