Page 60 - Undergraduate Research Opportunities Program Proceedings 2017-18

School of Engineering
Department of Civil & Environmental Engineering

DESR (Data-Enabled Scalable Research) Lab – Incubator of Creative Learners and Hub of Transformation

Supervisor: WANG Yu-Hsing / CIVL
Student: WEI Huan-ting / SENG Course: UROP1100, Summer

PyTorch is a powerful and popular deep learning tool based on one of the most popular language for machine learning,
Python. It is thoroughly introduced and taught in Professor Sung Kim’s online tutorial, which is the very beginning of my
learning of deep learning. In first part of this report I will focus on the basic concepts of deep learning including the
gradient descent, back propagation, and my own understanding toward it. I will also demonstrate building a simple
model with PyTorch by using linear layer, activation function, compute the loss with various loss functions, and optimize
the model with optimizers. Lastly, I will discuss about the idea and the structure of RNN (Recurrent Neural Network),
and one of its variations, the LSTM (Long Short Term Memory), of its approach to create a more accurate model. Deep
learning has taken us closer to the realization of artificial intelligence, and I would definitely like to learn more about it.

Improving Public Private Partnerships in Infrastructure Development

Supervisor: ZHANG Xueqing / CIVL
Student: DWIVEDY Sampriti / CIGBM Course: UROP1000, Summer

This UROP research report, follows research studies conducted by HKUST graduates in 2018 titled “Comparative Analysis
of International Practices of Public-Private Partnerships in Infrastructure Development and Management” for the United
States, Australia and United Kingdom. The objective of this report is to identify patterns in the development of PPP
legislations by comparing the historical development of PPP legislations in United States, Australia and United Kingdom.
The identified patterns can be learnt from and implemented in developing countries that are looking to adopt PPP
legislations.

An Analytical and Numerical Investigation of Debris Flow and Its Impact to Hong Kong

Supervisor: ZHAO Jidong / CIVL
Student: CHOI Chung Yin / CIVL Course: UROP1100, Spring
CHOW Tsz Yin / CIVL UROP1100, Spring

The particle size distribution is one of the main parameters that we need to focus on in our debris flow model design.
Different size distributions of particles can have different properties and effects in the debris flow. Since a lot of models
which are used to describe the debris flow assume that the particle size is the same for simplifying the calculation.
However, there are a range of size distributions in the actual situation. Therefore, we need to consider the particle size
distribution for the more accurate measurement and prediction for the real-life situation. Numerical simulation with
CFD-DEM provides aid for the further discussion and analysis.

Computer Visualisation of Sand Grains
Supervisor: ZHAO Jidong / CIVL
Student: LEUNG Ho Yin / CIVL Course: UROP1000, Summer

Soil mechanics is an important aspect in civil engineering. In the past, researcher usually do experiment with soil to
study its macroscopic properties. However, it was difficult for researchers to investigate the microscopic behavior of
sand grains. Nowadays, computer become more powerful than before. With the aids of computer, we can do massive

50