Undergraduate researchers from the Department of Mathematics and Statistics, under the supervision of Prof. Yoon Jinhee, have achieved consecutive awards at major AI conferences.

Lee Youngbin (Computer Engineering, ’23) and Joo Junhyung (Computer Engineering, ’24) received the Best Presentation Award at the 2025 International Symposium on Advanced Intelligent Systems (ISIS 2025), co-hosted by the Korea Intelligent Information Systems Society (KIIS) and the Japan Society for Fuzzy Theory and Intelligent Informatics (SOFT). The symposium, held at Chungbuk National University from November 6 to 9 under the theme “Reaching Beyond the Horizon,” featured approximately 160 presentations on advanced intelligent technologies, including generative AI, autonomous robotics, and smart factories.

The two students were recognized for their paper, “Soft Probability Flooring for Stable Learning: Revisiting Cross-Entropy under Outliers,” which proposes a novel cross-entropy variant to enhance learning stability in the presence of outliers. The study drew significant attention from participants. Notably, Joo further extended the research by developing and releasing a Python package, “robustloss-lab,” demonstrating both practical applicability and theoretical contribution.

In addition, Ko Seungho (Content Software, ’24) and Lee Hyunseok (Data Science, ’23) were selected for an Outstanding Paper Award at the 2025 Fall Conference hosted by the Korea Intelligent Information Systems Society, held at Seokyeong University from November 27 to 29. The conference featured oral and poster presentations across AI fields, including deep learning, machine learning, and autonomous driving, and included special sessions and tutorials for early-career researchers.

Their awarded paper, “Logarithmic Scale Weighted Cross-Entropy for Class-Imbalanced Learning,” introduces a logarithmic scale weighted cross-entropy loss function that effectively improves classification performance in severely imbalanced settings. The work was highly evaluated for its theoretical rigor and practical applicability.

Prof. Yoon Jinhee noted that integrating foundational mathematics with AI research across disciplinary boundaries has broadened students’ perspectives and led to tangible research outcomes, adding that the awards are especially meaningful as they stemmed from students’ voluntary engagement in research following academic advising.

Undergraduate researchers from the Department of Mathematics and Statistics, under the supervision of Prof. Yoon Jinhee, have achieved consecutive awards at major AI conferences.

Lee Youngbin (Computer Engineering, ’23) and Joo Junhyung (Computer Engineering, ’24) received the Best Presentation Award at the 2025 International Symposium on Advanced Intelligent Systems (ISIS 2025), co-hosted by the Korea Intelligent Information Systems Society (KIIS) and the Japan Society for Fuzzy Theory and Intelligent Informatics (SOFT). The symposium, held at Chungbuk National University from November 6 to 9 under the theme “Reaching Beyond the Horizon,” featured approximately 160 presentations on advanced intelligent technologies, including generative AI, autonomous robotics, and smart factories.

The two students were recognized for their paper, “Soft Probability Flooring for Stable Learning: Revisiting Cross-Entropy under Outliers,” which proposes a novel cross-entropy variant to enhance learning stability in the presence of outliers. The study drew significant attention from participants. Notably, Joo further extended the research by developing and releasing a Python package, “robustloss-lab,” demonstrating both practical applicability and theoretical contribution.

In addition, Ko Seungho (Content Software, ’24) and Lee Hyunseok (Data Science, ’23) were selected for an Outstanding Paper Award at the 2025 Fall Conference hosted by the Korea Intelligent Information Systems Society, held at Seokyeong University from November 27 to 29. The conference featured oral and poster presentations across AI fields, including deep learning, machine learning, and autonomous driving, and included special sessions and tutorials for early-career researchers.

Their awarded paper, “Logarithmic Scale Weighted Cross-Entropy for Class-Imbalanced Learning,” introduces a logarithmic scale weighted cross-entropy loss function that effectively improves classification performance in severely imbalanced settings. The work was highly evaluated for its theoretical rigor and practical applicability.

Prof. Yoon Jinhee noted that integrating foundational mathematics with AI research across disciplinary boundaries has broadened students’ perspectives and led to tangible research outcomes, adding that the awards are especially meaningful as they stemmed from students’ voluntary engagement in research following academic advising.

Reported by 권상혁 홍보기자(seankweon@naver.com)