2025/09/19 by Hideo Hirose, Hirose, Hideo
Computer Science · #Neural Networks and Applications
paper · pdf · doi:10.48550/arxiv.2509.15605
Checkered patterns are characterized by their square structure and the use of only two distinct colors. These colors are typically represented by two types of numerical sets: 1,0 and 1,-1. Matrices based on 1,0 may seem identical to those based on 1,-1 when forming checkered patterns because the only difference is that the numbers 0 are changed to -1. However, these two kinds of matrices are completely different in a mathematical sense because a matrix using 1,0 has a rank of 2 and a matrix using 1,-1 has a rank of 1. Knowing this difference in advance allows us to reduce the computational effort required for matrix operations such as information embedding and retrieving.