Here are ongoing research projects.

AI Platform to Fully Adapt and Reflect Privacy-Policy Changes Recent advancements in machine learning (e.g., deep neural network) exhibit varying applications. On the other hand, many voices raise a privacy concern, establishing a privacy policy accordingly such as GDPR (General Data Protection Regulation). The policy can be updated to catch up ever-changing technologies. This project aims to build an adaptable AI platform that is able to comply with an up-to-date privacy policy. This is a joint project with Simon Woo at Sungkyunkwan University (SKKU).

A Generalizable and Continual Deep Learning Model for Inferring the Context of Binary Codes This is a succeeding project to binary code representation with BERT. In particular, we explore a better metric to describe a binary property for a downstream task. Then, we study how to build a model for contextual inference of a binary in a generalizable and continual manner. Supported by NRF (National Research Foundation of Korea).

Past Projects

Code Unpacking Intermediate Representation Conversion This project focuses on the SGN packer, attempting to unpack a hidden code and to convert it to intermediate representation (IR) for further reliable execution. Supported by NSRI (National Security Research Institute).

A Metric to Measure the Quality of Decompiled Codes The project studies a metric (means) to evaluate quality of decompiled codes from varying decompilers such as Hex-Rays Decompiler. It encompasses an existing methodology for code quality measurement, comparison between an original source and a decompiled code, and tool development for measuring code quality. Supported by NSRI (National Security Research Institute). This is a joint project with Sungjae Hwang at Sungkyunkwan University (SKKU).

Efficient Fuzzing for Internal Communication Protocols with Firmware of Unmanned Vehicles The objective of the project is to discover a vulnerability against a protocol for internal communications in a Drone system (i.e., firmware binary). In particular, we look into protocols like uORB and FastRTPS in PX4, followed by investigating efficient fuzzing techniques that focus on a target protocol. This is a joint project with Daehee Jang at Sungshin Women’s University.

Semantic-aware Executable Binary Code Representation and its Applications with BERT. The project seeks a better binary code representation in an executable binary that can deduce the underlying code semantics with one of state-of-the-art architectures, BERT (Bi-directional Encoder Representations from Transformers). Further, we attempt to apply the representation to other applications that require the inference of code semantics. Supported by NRF (National Research Foundation of Korea).

Autonomous Car Security as part of Advance Research and Development for Next-generation Security. It aims to discover a vulnerability (e.g., sequence of malicious messages) against a car infortainment system, and to develop an AI-based intrusion detection system. Supported by IITP (Institute of Information & Communications Technology Planning & Evaluation). This is a joint project with Yuseok Jeon at Ulsan National Institute of Science and Technology (UNIST), Haehyun Cho at Soongsil University, and Dokyung Song at Yonsei University.

Vulnerability Analysis on IoTivity Protocol Provisioning. The goal of the project is to analyze the IoTivity protocol, an open source software framework for seemless communications between the Internet of Things as an implementation of OSF (Open Software Framework), followed by discovering a weakness of the protocol while onboarding and provisioning. Supported by NSRI (National Security Research Institute). This is a joint project with Hyongkee Choi, and Jaehoon Paul Jeong at Sungkyunkwan University (SKKU).