RGC Senior Research Fellow Scheme

Biography

Choh-Ming Li Professor of Computer Science and Engineering in Department of Computer Science and Engineering at The Chinese University of Hong Kong
Research interests include software engineering, dependable computing, distributed systems, cloud computing, mobile networking, big data, and machine learning
SRFS project — aims at investigating Large Language Models (LLMs) on three main scopes: evaluation, exploration, and application of LLMs in the context of code intelligence. The project will create the first comprehensive platform to evaluate LLMs for code production, explore LLM development and employment stages in practical Software Engineering processes, and apply LLMs in real-world software development processes for fully automated software reliability engineering
Awards and Honours:
- RGC Senior Research Fellow (2024)
- HKAES Fellow (2021)
- The 13th Guanghua Engineering Science and Technology Award (2020)

Project Title

Evaluation, Exploration, and Application of Large Language Models on Code Intelligence

Award Citation

Professor Michael Rung-Tsong Lyu is the Choh-Ming Li Professor of Computer Science and Engineering at The Chinese University of Hong Kong (CUHK). In the early 1990s, he initiated the research area of Software Reliability Engineering (SRE) by co-founding the first international symposium in SRE and publishing seminal SRE books. His contributions led him to the elections of IEEE Fellow (2004), AAAS Fellow (2007), IEEE Reliability Society Engineer of the Year (2010), and ACM Fellow (2015).

Professor Lyu started to investigate on Machine Learning and AI in 2010. He and his colleagues developed new machine learning models for social recommendation, resulting in three Test-of-Time Paper Awards 10 years later. His team also launched the LogPAI (Log Analytics Powered by AI) project, which is highly influential in both industry and academia.

Professor Lyu currently devotes his research on the intersection of AI and Software Engineering (SE). Recognizing the new software engineering paradigm offered by Large Language Models (LLMs), in this research project, Professor Lyu aims at unleasing the power of LLM for software code generation, minimizing its development cost, and maximizing its reliability. The project will focus on three main scopes: evaluation, exploration, and application of LLMs in the context of code intelligence. It will create the first comprehensive platform to evaluate LLMs specifically for code production. This evaluation investigates LLM’s language comprehension, knowledge, reasoning, robustness, and security abilities, and explore how well LLMs perform on various code-related tasks. Furthermore, the project will explore LLM development and employment stages, including pre-training, fine-tuning, prompt learning, and post-processing, in order to enhance the quality and generalizability of LLMs in practical SE processes. Finally, the project will apply LLMs in real-world software development processes, assesses their roles in SE lifecycles, and integrates them for fully automated SRE.