Including the detection of pishing attacks, Domain Generation Algorthims (DGAs), and Software Defined Networking (SDN)
To understand how phishing emails will affact people's behavior, we design a set of experiments to study the user behavior encoutered with phishing attacks
In most of my researches, I explore machine learning techniques
I mainly worked on the physical based simulation and deformable objects with level of details
To recognize a certain facial expression based on Action Units (AUs) detection by using Kinect 2 and thermal camera. We use machine learning techniques to train the AUs for the automatic action units detection
I mainly worked on multi-scale image decomposition with the purpose of reproducing the real scene more realistic. In order to display a High Dynamic Range Image (HDR) in traditional output devices, we use a multi-scale tone mapping approach by using differet filter
In this project, in order to efficiently avoid waste of CPU cycles, we have implemented five different branch predictors. Additionally, we have implemented a parser that used to fetch instruction address and branch result from a standard trace file. We test our predictors by using different trace files which are selected from SPEC 2000. As a conclusion, the size of trace file has dramatically influence on the prediction accuracy.
In this project I build an augment reality system that provides the function that allow people to interact with a virtual sphere and particle system (applied with fire shader and star shader). My system was designed to use Kinect 2 to capture the human movement. The user interface of my system was build using Qt to change from different models.
In this project, we built a scheduling system for a hard disk. We created a virtual disk object and a simple program that get requests from that disk. Our scheduling system uses the implemented simulators to deal with the requests and get disks access time. Then we compared the performance between each implemented simulators.
I came up with my an algorithm for solving art gallery problem. Specifically, first, I calculate the visibility polygon and find the atomic visibility polygons (AVPs), then add those AVPs to each vertex visibility polygon. Second, I reduce this problem to the Set Cover Problem, which can be solved using approximate approach. My proposed method reduces the runtime of polygon discretization using AVPs(O(n^3)) compared with other approximate method using convex component(O(n^4)).