Huang Zhiying, Hosei University, 2nd year of Master’s course of Computer and Information Sciences, was awarded IEEE Outstanding Paper Award at the 1st International Workshop on Generative AI and Hyper Intelligence (GAI-HyperI 2024) at 9th IEEE Cyber Science and Technology Congress (CyberSciTech 2024).

受賞情報

学会名/Conference Name

The 1st International Workshop on Generative AI and Hyper Intelligence (GAI-HyperI 2024) at 9th IEEE Cyber Science and Technology Congress (CyberSciTech 2024)

学会開催期間/Conference Period

2024年11月5日 ～ 2024年11月8日/5-8 Nov 2024

発表日および受賞日/Presentation and Awarded Date

2024年11月6日/6 Nov 2024

会場/Conference Venue

Boracay Island, Malay, Philippines

論文題目/Paper’s title

Leveraging ECG Signal for People Identification under Different Emotion States

著者/Authors

• Huang Zhiying, 法政大学情報科学研究科修士課程2年/Hosei University, 2nd year of Master’s course of Computer and Information Sciences
• Yuang Meng, 法政大学情報科学研究科修士課程1年/ Hosei University, 1nd year of Master’s course of Computer and Information Sciences
• Walid Brahim, 法政大学情報科学研究科博士後期課程2年/ Hosei University, 2nd year of Doctoral course of Computer and Information Sciences
• Ao Guo, 名古屋大学情報学研究科研究員/Nagoya University, Researcher of Graduate School of Informatics, Nagoya University
• Jianhua Ma, 法政大学情報科学部教授/Professor of Hosei University, Faculty of Computer and Information Sciences

Different emotional states have been shown to affect the accuracy of people identification based on physiological signals. However, current studies have not fully clarified the impact of various emotion types, leaving it uncertain whether people identification can consistently maintain accuracy under different emotional states. The CASE dataset includes physiological signals collected from different participants while watching different types of videos. These videos are designed to evoke different emotions (i.e., amusing, boring, relaxed, and scary). We assume that participants are in different emotional states while watching different types of videos. ECG signal were used to build a 1D-CNN model for people identification. To clarify the performance of people identification in different emotional states, we first determined the optimal window and sliding sizes for data processing across all emotional states. We then tested the optimal pool size and duration to achieve promising identification accuracy. By comparing the performance of the identification model under different states, we observed that ECG-based people identification achieved promising accuracy, indicating its potential to be effective in various scenarios when individuals are experiencing different emotions.