The Impact of Large Language Models: From Perception Computing to System Intelligence

Prof. Michael Lyu, CUHK

Date: 22 September 2023 (Friday)
Time: 3:20-3:40pm HKT
Venue: IAS4042 Seminar Room, Lo Ka Chung Building, HKUST
Title: The Impact of Large Language Models: From Perception Computing to System Intelligence
Speaker: Prof. Michael Lyu, CUHK

Abstract

Large Language Models (LLMs) have become increasingly popular in recent years, with models like ChatGPT and GPT4 leading the charge. These models have demonstrated remarkable capabilities in generating human-like text, making them invaluable tools in a variety of applications, from drafting emails to writing code. This talk starts with an introduction to LLMs and their impact, with a special focus on Natural Language Processing (NLP) and Computer Vision (CV) tasks. Then, we talk about the exploration of our work on the application of LLMs in Artificial Intelligence Development (AIDev), featuring the use of prompt tuning and in-context learning in code intelligence. We will also delve into the domain-aware prompt-based type error repair, demonstrating how LLMs facilitate research in this area. The talk then includes the application of LLMs in Artificial Intelligence Operations (AIOps), demonstrating our work on scalable and adaptive log anomaly detection, fault-indicating information extraction in logs, and LLM-powered logging statement generation. These applications aim to explore the potential of LLMs to improve AIOps by leveraging the abilities of LLMs to understand the semantics of domain-specific languages (e.g., logs). Lastly, we address the robustness of LLMs, focusing on the challenges and potential solutions related to toxicity, social bias, and cultural bias in these models. This section aims to shed light on the ethical considerations in the development and deployment of LLMs, underscoring the importance of creating models that are not only technically proficient but also socially responsible. Overall, this talk aims to provide a holistic understanding of the impact of LLMs, from perception computing to system intelligence, and to inspire further exploration and innovation in this rapidly evolving field.

Biography

Michael R. Lyu received his Ph.D. in Computer Science from the University of California, Los Angeles, USA. He is currently Choh-Ming Li Professor of Computer Science and Engineering at The Chinese University of Hong Kong. Prof. Lyu's research interests include software engineering, software reliability, machine learning, cloud and mobile computing, and distributed systems. He has published over 600 refereed journal and conference papers in his research areas. His Google Scholar citation is over 51,000, with an h-index of 110. Prof. Lyu initiated the first International Symposium on Software Reliability Engineering (ISSRE) in 1990. Prof. Lyu was elected IEEE Fellow (2004), AAAS Fellow (2007), ACM Fellow (2015), HKAES Fellow (2021), and named IEEE Reliability Society Engineer of the Year (2010). He was granted the second-class award in Natural Sciences from the Ministry of Education, CCF Overseas Outstanding Contributions Award in 2018, and the 13th Guanghua Engineering Science and Technology Award in 2020. He was also named in The AI 2000 Most Influential Scholars Annual List in 2020.