Human Behavior Inference via Analyzing Record-and-Replay Logs
Human-in-the-loop has been an extremely important factor to consider in cybersecurity analysis. For example, most APTs start with social engineering to help the adversary gather target information. However, it is a difficult problem remaining unsolved, mainly due to the fact that human behaviors are hard to predict and trace. The purpose of this project is to develop a novel technique to infer such non-deterministic user/human behaviors. We plan to use a record-and-replay system which can faithfully reproduce the execution of the whole system and program executions by lightweight logging for data collection, and relationship inference and causal analysis techniques inspired by the existing artificial intelligence community for human behavior inference. This work is done under direction of Prof. Shiqing Ma.
Trustworthy Data-Driven Framework for Location Based Online-to-Offline Commerce
Developing a trustworthy management framework for LBO2O commerce platforms is essential for ensuring social inclusion and equity of future O2O workers in the era of AI because the impacts of biased decisions are likely to be concentrated in some sub-communities of workers, leading to the social inequality. This work was advised by Prof. Mary Rigdon, in relation to an NSF application.
Read my survey summary
RPL7 Silencing Effect on Neurite Development
This study is intended to continue ongoing experiments concerning RPL7 and its role in neocortical development in the embryo. Specifically, this experiment examines neuron morphology between control groups and RPL7 knockdown groups at DIV4 and DIV8. This study was done under direction of Prof. Mladen-roko Rasin.
Read my work report