1. Full name: Dao Thanh Tung                                        2. Sex: Male

3. Date of birth: November 5th , 1975                             4. Place of birth: Ha Tinh

5. Admission decision number: No 4438/QĐ-ĐHKHTN, dated on November 26th , 2015 by Rector of VNU University of Science.

6. Changes in academic process:

Extension decision number 2402/QĐ-ĐHKHTN, dated on August 02nd, 2019 by Rector of VNU University of Science.

7. Official thesis title: Selection of variables, component numbers and parameter estimates by VB method for GLMM and MRDE-MN models.

8. Major: Theory of Probability and Mathematical Statistics.

9. Code: 9460112.02

10. Supervisors:

            Assoc.Prof.Dr. Tran Minh Ngoc

            Dr. Tran Manh Cuong

11. Summary of the new findings of the thesis

- The thesis proposes a new, fast algorithm to perform simultaneous selection of variables and parameter estimation for GLMM model in case of multi-variables.

- The thesis is the first work to study the MRDE-MN model and propose the algorithm to perform three objectives simultaneously including variable selection, componential number determination and parameter estimation for MRDE-MN model.

The algorithms proposed in the thesis have been evaluated as having good performance through testing on simulation data and real data.

12. Practical applicability, if any:

- The thesis has built two fast algorithms with good performance, which can be applied in practice on real data with the GLMM model and the MRDE-MN model.

- GLMM and MRDE-MN models in High-dimensional are difficult to solve, but the two algorithms built in the thesis have good performance and high efficiency.

13. Further research directions, if any

- Extend the method for grouped variable selection in GLMMs by using the group lasso penalty.

- Extend the method for ordered variable selection in GLMMs by the composite absolute penalty.

- Extend the method for variable selection and component selection in the mixture of experts model where each component is a neural network.

14. Thesis-related publications:

Dao Thanh Tung, Minh-Ngoc Tran, Tran Manh Cuong, Bayesian Adaptive Lasso with Variational Bayes for Variable Selection in High-dimensional Generalized Linear Mixed Models, Communications in Statistics – Simulations and Computations, 0, 1 – 14, 2018.

DOI: 10.1080/03610918.2017.1387663.

Dao Thanh Tung, Minh-Ngoc Tran, Flexible Multivariate Regression Density Estimation, Communications in Statistics – Theory and Methods, 2020, DOI: 10.1080/03610926.2020.1723633.