-
Hiroshi KERA
Researchmap
Tenure Track Associate ProfessorChiba University Institute for Advanced Academic Research / Graduate School of Informatics
Keywords
Computational Algebra, Deep Computational Algebra, Computer Vision
Professional Memberships
IEEE Computational Intelligence Society, Japan Society for Symbolic and Algebraic Computation
Research Theme
Learning-driven computation in algebra and robust computer vision

Abstract
Large-scale algebraic systems arise in a wide range of applications, including dynamical systems analysis, computer vision, and post-quantum cryptography. However, computational-algebraic problems, including solving algebraic systems, often involve extremely high computational complexity, making it difficult to solve problems at the scale required for practical applications.
I aim to breakthrough these limitations from a new angle using deep learning, and am pioneering new fields of "Computational Algebra for Learning" and "Deep Learning for Computation." More generally, I am interested in deep learning approaches for transformations that are highly sensitive to input changes, such as algebraic computations. Through broad application research in algebra, vision, meteorology, and other fields, I am developing unified methodological and theoretical foundations for deep learning that can handle various forms of input sensitivity.