SEOUL, May 18 (AJP) - An undergraduate student in South Korea has authored a research paper on artificial intelligence that has been accepted for presentation at the 43rd International Conference on Machine Learning, Kookmin University said Monday.
Kim Min-woo, a senior in the university's software program, is the first author of the paper titled "Memory as Dynamics: Learning Reliability-Guided Predictive Models for Online Video Perception." The research introduces a framework for online video perception that interprets memory as a dynamic system rather than a static storage unit. By applying a reliability-guided predictive model, the system processes temporal information within video sequences to improve accuracy and efficiency.
The model dynamically estimates the reliability of each video frame and incorporates this assessment into its memory updates and predictions. This process allows the system to maintain stable recognition performance even when the video feed contains noise or visual obstructions. The method demonstrated improved performance across various online video benchmarks when compared to existing techniques.
The framework has potential applications in fields that rely on real-time visual recognition, including autonomous driving, robotics and intelligent video understanding. Kim's research was supported by the National Research Foundation of South Korea (NRF) and the Institute of Information and Communications Technology Planning and Evaluation (IITP).
"I wanted to untangle the relationship between memory and prediction from a new perspective," Kim said. "It is very meaningful that the research I conducted during my undergraduate studies has been recognized for its achievements at a world-class academic conference. I want to continue practical AI research that can contribute to solving real-world problems."
Copyright ⓒ Aju Press All rights reserved.
