세미나 소개
---------------------------------------------------------------------------
- 주제: Hypercomplex-valued Neural Networks with Applications
- 연사: Dr. Agnieszka Niemczynowicz, Dr. Radoslaw Kycia
- 소속: Carow University of Technology, Faculty of Computer Science and Mathematics
- 일시: 2026년 4월 28일 (화) 오후 2:00 - 3:00
- 장소: 한양대학교 FTC 522호
- 주관: 산업AI부트캠프사업단, 인공지능대학원
세미나 주제:
- The importance of mathematical foundations and analytical properties in AI
- Introduction to basic concepts and theoretical foundations of Hypercomplex-values Neural Networks (HvNNs)
- Key Mathematical and methodological challenges in this area, including issues related to model design training, and interpretation
- The role of hypercomplex algebras in shaping the behavior and capabilities of HvNNs
Abstract:
Hypercomplex-valued neural networks (HvNNs) constitute a rapidly developing branch of machine learning that extends classical neural architectures to higher-dimensional algebraic systems, including complex numbers, quaternions, and octonions. By operating in hypercomplex domains, these models offer a mathematically rich framework for representing multidimensional data, preserving intrinsic structural relations, and exploiting parameter-sharing mechanisms that can lead to more compact network representations.
This lecture provides an overview of the theoretical foundations of hypercomplex-valued neural networks, with particular emphasis on their underlying algebraic structures, representation principles, and selected architectural formulations. The talk also discusses key mathematical and methodological challenges arising in this area, including issues related to model design, training, and interpretation. Special attention is given to the role of hypercomplex algebras in shaping the behavior and capabilities of these networks.
Overall, the lecture aims to highlight the theoretical significance of HvNNs as an emerging research direction at the intersection of mathematics and artificial intelligence, while indicating the need for further rigorous development of their foundations and analytical properties.
문의: 한양대학교 산업AI부트캠프사업단 (yoonseonchoi@hanyang.ac.kr)