Abstract: To address the problem of radius sensitivity in granular ball based spatial partitioning, which results in coverage gaps or region overlaps, an information granulation method is proposed based on n-dimensional hypercubes. The traditional constraint of spherical structures is overcome by introducing a novel n-dimensional hypercube geometric model, which establishes a theoretical framework for spatial partitioning without coverage gaps or overlapping regions. A fast granular hypercube generation(FGHG)algorithm is proposed, which utilizes a dimension-adaptive partitioning mechanism to enable efficient spatial division. Compared with traditional granular ball generation algorithms, FGHG algorithm demonstrates significant advantages in computational efficiency. A fast granular hypercube classifier(FGHC)is designed. To validate the effectiveness of the proposed algorithm, a systematic evaluation is conducted on 13 real-world datasets from the repository, where FGHC algorithm achieving improvements in both classification accuracy and F1 score. The granular hypercube computing paradigm established in this study provides a novel theoretical framework for tackling complex spatial partitioning problems in data analysis.

Key words: granular computing, granular hypercube, information granulation, classifier, granular ball computing