Download 附件1：内容简介： 报告一：Introduction to deep learning 报告人

附件 1：内容简介：
报告一：Introduction to deep learning
报告人：Zenglin Xu
Abstract:
In this talk, I will give a general introduction to deep learning. It covers current popular deep
learning models, including CNN, RNN, and deep reinforcement learnig. I will also review current
research directions and open problems to deep learning.
报告二: Partial Projective Resampling Method for Dimension Reduction: With Applications to
Partially Linear Models
报告人：Wenbo Wu
Abstract:
In many regression applications, the predictors naturally fall into two
categories: “the predictors of primary interest” and “the predictors of secondary
interest”. It is often desirable to have a dimension reduction method that focuses on
the predictors of primary interest while controlling the effect of the predictors of
secondary interest. To achieve this goal, a partial dimension reduction method via
projective resampling of a composite vector containing the response variable(s) and
the predictors of secondary interest is proposed. The proposed method is general in
the sense that the predictors of secondary interest can be quantitative, categorical or a
combination of both. An application of the proposed method for estimation in
partially linear models is emphasized. The performance of the proposed method is
assessed and compared with other competing methods via extensive simulation. The
empirical results show that, in addition to the superior estimation accuracy, the
proposed method has a considerable computational advantage. We also demonstrate
the usefulness of the proposed method by analyzing two real datasets.
报告三：Dimension reduction for multivariate spatial data
报告人： Qin Wang
Abstract:
Dimension reduction provides a useful tool for analyzing the high dimensional data. The recently
developed Envelope method is a parsimonious version of the classical multivariate regression
model. However, existing envelope approaches do not address the additional complications
associated with spatial or temporal correlations in some real applications. Motivated by two data
sets (from brain imaging and environmental studies respectively), we combine the ideas of the
envelope method with multivariate spatial statistics to introduce a new approach, Spatial Envelope.
This approach provides efficient estimates for the parameters of interest while being able to
capture the spatial structure in the data.
报告四: Homogeneity Pursuit in Single Index Models based Panel Data Analysis
报告人：Wenyang Zhang
Abstract:
Panel data analysis is an important topic in statistics and econometrics. Traditionally, in panel data
analysis, all individuals are assumed to share the same unknown parameters, e.g. the same
coefficients of covariates when the linear models are used, and the differences between the
individuals are accounted for by cluster effects. This kind of modelling only makes sense if our
main interest is on the global trend, this is because it would not be able to tell us anything about
the individual attributes which are sometimes very important. In this talk, I will present a new
modelling approach, based on the single index models embedded with homogeneity, for panel
data analysis, which builds the individual attributes in the model and is parsimonious at the
same time. I will show a data driven approach to identify the structure of homogeneity, and
estimate the unknown parameters and functions based on the identified structure. I will show the
asymptotic properties of the resulting estimators. I will also use intensive simulation studies to
show how well the resulting estimators work when sample size is finite. Finally, I will apply the
proposed modelling idea to a public financial dataset and a UK climate dataset, and show some
interesting findings.