Accepted and published work is listed in full. Papers still under review or in preparation are shown as in progress, and updated with venue and links once accepted.
I. Lee
Dynamic DNN partitioning and cooperative offloading across heterogeneous devices on EV-Hub edge infrastructure, optimizing latency and energy for public-safety networks. Accepted; forthcoming 2026.
S. Kwon
A methodological study of when and how field experiments can be applied to evaluate foreign police-assistance programs — the design challenges they raise and the causal-evidence opportunities they offer.
I. Lee
Analyzes the trade-offs policing AI faces among competing goals using an LLM-based synthetic expert panel, and derives four future scenarios through 2×2 scenario planning. Accepted for publication.
I. Lee
A time layer that keeps drones and robots operating on trustworthy time under spoofing, for public-safety physical AI. To be presented on October 1, 2026.
Active research across our four programs. Details are added as each paper is accepted for publication.
I. Lee
I. Lee
I. Lee
I. Lee
S. Kwon
Korean applications, listed by title and number. Technical details are withheld until each application is published.
English title (unofficial): Self-reconfiguring distributed inference method and system. Korean patent application, examination requested. Applicant: KAIST.
English title (unofficial): System and method for identifying and tracking the actions of autonomous AI agents using active cognitive stimuli and digital antigens. Korean provisional application. Applicant: KAIST.