NYCU Freshman Yu-Lin Tsai’s Paper Accepted by Top AI Academic Conference NeurIPS

  • Post category:News
From right to left: Yu-Lin Tsai, Chia-Yi Hsu, Asst. Prof. Chia-Mu Yu, and Pin-Yu Chen

A research paper titled “Formalizing Generalization and Robustness of Neural Networks to Weight Perturbations” coauthored by members of National Yang Ming Chiao Tung University (hereafter NYCU) has recently been received by NeurIPS, a top conference in the artificial intelligence (AI) field, and much attention has been given to the first author in particular, Yu-Lin Tsai, a freshman at the Arete Honors Program of NYCU. The other coauthors are Chia-Yi Hsu, a PhD student at the Department of Computer Science; Chia-Mu Yu, an assistant professor at the Department of Information Management and Finance; and Pin-Yu Chen, an alumnus at EECS Undergraduate Honors Program.

NeurIPS, a top academic conference on topics related to machine learning and computational neuroscience, is the world’s largest academic conference on AI and the most significant international event in the AI research field. NeurIPS ranks second in the AI category of Google Scholar and has an overall ranking of 12th in the natural science category. Tsai earns high praise from the conference for his collaboration—as a freshman at the NYCU Arete Honors Program and the first author—with members from other NYCU Departments, such as the Department of Information Management and Finance and the Department of Computer Science, on research into methods for enhancing machine learning model stability.

Tsai scored 970 at TOEIC, which is nearly the perfect score, when he was a sixth grader; studied in the science class of Taichung Municipal Taichung First Senior High School; was the first senior high school student admitted to Taiwan AI Academy; and won the 1st place in the final project presentation. With these achievements, Tsai was qualified to be admitted to related departments at almost all universities in Taiwan. When asked about the reason for his choice of the NYCU Arete Honors Program, Tsai said that the program offers great flexibility and freedom to students in its course selection requirements. The program’s easing restrictions on course selection allows him to engage in explore self-exploration and interdisciplinary learning without being limited to the traditional framework of course credits and required/elective courses. Tsai is currently taking courses at the Department of Computer Science and Department of Applied Mathematics and concurrently doing research with graduate students at the laboratory of assistant professor Chia-Mu Yu at the Department of Information Management and Finance.

Jiun-Yu Wu, a professor at the Arete Honors Program, said that NYCU has been breaking down boundaries between departments since the 2018–2019 academic year and introduced the Arete Honors Program, the first four-year program that is not limited to any discipline and admits students based on their special achievements. Students admitted to the program are allowed to take technical courses across all colleges; such flexible and autonomous way of learning is expected to foster interdisciplinary and international talent in the next generation. According to his observation, professor Wu said that the program is suitable particularly to students like Tsai, who has extraordinary performance in the information field, published two first-class papers in international conferences when he was at the twelfth grade in senior high school, enjoys communicating with others, and is interested in entrepreneurship and even philosophy. Take AI, a field Tsai is interested in, for example: For AI to be applied in more valuable ways, understanding of the users—which requires knowledge in psychology and philosophy—is needed on top of the fundamental machine computation.

The Arete Honors Program has allocated resources to and made several investments in students with special expertise in NYCU through the provision of an interdisciplinary program, a flexible credit policy, project cocreation courses, academic forums, technical and practical courses—including off-campus practicum—and a coworking space. The program has admitted four cohorts of students, who share the trait of having strong intrinsic motivation and time management skills. They all create an interdisciplinary autonomous learning map of their own, ready to make the most out of their four years at NYCU.

Chinese version: