Professor Kim Joong-heon and Professor Lee Sang-keun selected for the 2024 SW Star Lab project
Professor Kim aims to develop ultra-low-cost quantum AI-based second-life platform technology.
Professor Lee seeks to overcome the inference limitations of current large language models (LLMs).
Both professors will receive support for up to eight years in the field of core software technology.

▲ Professor Kim Joong-heon from the School of Electrical Engineering (left), Professor Lee Sang-keun from the Department of Artificial Intelligence (right)

The labs selected for this project will receive support for reliable research for up to eight years depending on research performance.

The SW Star Lab initiative, launched by the Ministry of Science and ICT in 2015, aims to train talented master’s and doctoral students who will lead international software innovation while securing world-class foundational and source technologies in core software fields. Labs selected for this project will receive reliable support for research for up to eight years depending on their research performance.

Of the 10 labs selected, Korea University was chosen in two fields: big data and cloud computing.

In the cloud fieldIn association with cloud computing, Professor Kim aims to develop ultra-low-cost quantum AI-based second-life platform technology. Virtual environments, such as digital twins and metaverses, create spaces where people and objects interact, merging reality and virtuality to generate new value. These environments are gaining attention for their ability to overcome physical distance constraints and provide innovative experiences and services to users.

Unlike existing social platforms, where physical and virtual spaces are often blurred, virtual environments clearly distinguish between the two. However, research in this area remains insufficient. Additionally, current algorithms struggle with the high computational demands and complexities of virtual environments, making real-time computation challenging. To address these issues, a new algorithm is needed to ensure stable real-time performance and efficient data processing in virtual environments.

Professor Kim's research focuses on utilizing a quantum artificial intelligence (AI) algorithm that offers benefits such as dimensionalityreduction, high-speed computation, and more lightweight models. The goal is to provide ultra-fast, ultra-low-cost real-time services that consider the constraints of physical space within a virtual environment, ultimately developing quantum AI-based second-life platform technology.

Professor Lee is developing big data with the goal of “Karma: Advanced Inference through Knowledge Enhancement.” While LLMs have demonstrated remarkable performance in various fields, they still struggle with complex inference tasks that require reaching correct answers through multiple steps or considering multiple modalities, such as language, vision, and cultural context. They also fall short in areas such as common sense and ethical reasoning.

This project ultimately aims to overcome these limitations in current LLMs. To achieve this, Professor Lee plans to separate the reasoning capabilities and knowledge embedded in existing LLMs and design a new approach that naturally enhances their reasoning abilities with diverse knowledge from the world. Through this innovative method, he aims to surpass the limitations of current LLMs and ultimately attain human-level reasoning capabilities.