Summary
International Conference on Emerging Technologies for Communications
2023
Session Number:P2
Session:
Number:P2-34
Preliminary Study of the Importance of Small Eigenvalues through Matrix Approximation without Rank Restriction
Eriko Segawa, Yusuke Sakumoto,
pp.-
Publication Date:2023/11/29
Online ISSN:2188-5079
Summary:
The discussion of the low-rank approximation conducts that the large eigenvalues of a matrix contain more information than the small ones. Therefore, many graph algorithms use large eigenvalues. On the other hand, we clarified that the performance of a graph algorithm can be improved by using the combination of top and bottom eigenvalues. However, it has not been fully understood when and why the bottom eigenvalues are useful. In this paper, we first construct the matrix approximation without the rank restriction in the low-rank approximation and show that when bottom eigenvalues are useful.