Project Level: Winter

Project Duration:

4 weeks - 36 hours per week. Applicant will be required on-site for this project.


In recent years, deep artificial neural network is empirically proved able to solve complicated machine learning problems, including image recognition, speech analysis, natural language processing, nonlinear regression and classification, and so forth. This project aims to explore the power of deep learning on an important family of problems in statistics, the density estimation problem. In particular, we focus on the estimation of mixing density of Poisson models. This includes both count data and grouped-andright- censored categorical data. Because of the very short duration, this project will focus on empirical testing of the deep learning process.

Expected Outcomes:

1. Scholar will learn the deep learning platform PyTorch, and its operation on UQ high performance computing clusters (e.g., Tinaroo).
2. Scholar will learn the density estimation problem for the Poisson mixture models.
3. Scholar will learn basic concepts of statistical inference.

In particular, scholar will be invited to deliver a presentation at the end of the research program.

Suitable for:

This project is open to applications from students with a background in mathematics or statistics, 3rd year undergraduate program or above.
Because of the nature of empirical implementation in this project, applicant is expected to have strong interest in programming.

Project members

Dr Xin Guo

Senior Lecturer in Mathematical Dat
School of Mathematics and Physics