Ji-Hyeon Jeong
Hanyang University · 情報科学
研究室紹介
Professor Ji-Hyeon Jeong's research lab specializes in high-performance computing and intelligent systems, focusing on optimizing computational efficiency for complex engineering simulations and advancing the design of personal intelligent agents. The lab develops cutting-edge algorithms and frameworks for parallel computing—particularly on many-core CPUs and GPUs—enabling real-time simulation of dynamic systems such as flexible bodies and wheeled vehicles. Additionally, the lab investigates cognitive and behavioral aspects in large language model (LLM) agents, particularly the unintended influence of personal memory biases on decision-making, as seen in their work on memory-induced tool-drift. The research bridges low-level system optimization with high-level AI agent intelligence, emphasizing performance, scalability, and human-centered design.
Research Overview
Research Output Trend
Figures are computed from collected data and may differ slightly.
Selected Papers
6Model sizes have increased significantly in the fields of engineering and scientific computation. Some additional computing devicessuch as GPU, accelerators and co-processors have been applied to improve the computation performance. This paper presents severalstrategies to optimize the computation performance. The first strategy is to combine a computation unit with multiple of 4-tetrahedrons tosupport AVX vectorization. The second strategy is to utilize a GPU device. Several techniques are prop
Personal intelligent agents (IAs) are increasingly embedded in everyday life, a trend accelerated by generative AI technologies. Despite their growing presence, these agents often remain fragmented across different life domains and environments. This workshop explores how to design integrated IA ecosystems emphasizing continuity, coordination, and human-centered values. Participants with varied perspectives will collaboratively develop frameworks, scenarios, and guidelines for cohesive personal
This research proposes an effective implementation of linear equation solver for an implicit integration on a many-core CPU. Although this implementation is applied to a flexible body simulation in mechanical dynamics, it could be also utilized in a wide range of other fields. BFS-based nested dissection and its numerical factorization enables adaptive control of setting operational range as well as positive parallelization compared with traditional DFS-based nested dissection. It brings better
Modern LLM agents combine long-term memory for personalization with tool-calling interfaces for taking actions in the world -- a combination underpinning contemporary production systems. We study a previously unexamined failure of this combination: when personality-driven biases stored in memory (cost-consciousness, impatience, risk tolerance, etc.) silently affect tool calls in contexts where they are not applicable. We call this memory-induced tool-drift and operationalize it through MEMDRIFT,
Modern LLM agents combine long-term memory for personalization with tool-calling interfaces for taking actions in the world -- a combination underpinning contemporary production systems. We study a previously unexamined failure of this combination: when personality-driven biases stored in memory (cost-consciousness, impatience, risk tolerance, etc.) silently affect tool calls in contexts where they are not applicable. We call this memory-induced tool-drift and operationalize it through MEMDRIFT,
A realtime simulator using an explicit integration method is introduced to improve the solving performance for the dynamic analysis of a wheeled vehicle. Because a full vehicle system has many parts, the development of a numerical technique for multiple d.o.f. and ground contacts has been required to achieve a realtime dynamics analysis. This study proposes an efficient realtime solving technique that considers the wheeled vehicle dynamics behavior with full degrees of freedom and wheel contact