Page 77 - Undergraduate Research Opportunities Program Proceedings 2017-18

School of Engineering
Department of Computer Science & Engineering

problem of predicting stock prices. Without any professional experience, amateur investors and analysts may potentially
use this information to make decisions when performing trades.

Machine Learning and Anti-money Laundering

Supervisor: KWOK James Tin Yau / CSE
Student: WONG Ngo Yin / COMP Course: UROP1100, Spring
In this project, we used two different machine learning algorithms, support vector machine (SVM) and decision tree, to
classify which customers using the banking service are potentially laundering money. We compared the difference of
these two approaches in terms of their performance. The dataset used in this project is provided by an online bank in
China, with basic information of the customers and transaction records of their accounts. We first preprocessed the data
and designed features before applying our models.

Software AI agents
Supervisor: LIN Fangzhen / CSE
Student: ZENG Kuang / COMP Course: UROP1100, Spring

After AlphaGo defeated a series of best Go players, Artificial Intelligence has become the one of the hottest topics.
Similar with Go, Five-in-a-row is also an abstract strategy board game for two players, in which the aim is to put five
same pieces in a row more quickly than opponent. Compared to Go game, Five-in-a-row game has lower complexity and
simpler rule so the project study human-computer game reasoning based on Five-in-a-row game. Some search methods,
decision-making planning algorithm and game analysis will be discussed, varying from basic game tree to complex
machine learning and optimization methods. An intelligent Five-in-a-row agent is implemented according to mentioned
algorithms.

Multimodal Interaction with Sensor Fusion in VR/AR
Supervisor: MA Xiaojuan / CSE
Student: CHEN Yuan / COMP Course: UROP2100, Fall
LIANG Zhibo/ CPEG UROP2100, Fall

The ultrasonic sensor Welle outputs 2D x-y coordinate points when a person’s hands are approaching it. However, these
points could be easily polluted by the environment and needed to be mapped to gestures. Besides, in UROP 1100, the
VR system we proposed was lack of entertainment. To address these two issues, we propose the new VR system
“WellVR2”, which consisted of two parts, namely the server system and the VR mobile system. In this note, we will
introduce how we design and analyze the gesture-mapping and test gesture-mapping control, how to reduce the noise
and implement the system architecture. In the end, we will present an informal evaluation on the robustness of this
approach.

Air Quality Data Analysis and Prediction Using Machine Learning Approach

Supervisor: NG Wilfred Siu Hung / CSE
Student: CHA Seunghwan / CPEG Course: UROP1000, Summer
ROHATGI Tanay / COGBM UROP1000, Summer

67