In this blog post, the author discusses the concept of Singular Value Decomposition (SVD) and its applications in image compression. SVD is a matrix factorization technique that breaks down a matrix into three distinct matrices: $mathbf{U}$, $mathbf{Sigma}$, and $mathbf{V}^mathsf{T}$. The author explains the geometric interpretation of SVD and how it can be used to decompose a linear transformation into rotations and scaling. The author also demonstrates how SVD can be used to extract information from a given dataset by analyzing the singular values. In addition, the author explores the application of SVD in image compression, showing how a cat image can be reconstructed using only the most important components. The author concludes by highlighting the usefulness of SVD in various fields, including machine learning, finance, and data science.
https://dmicz.github.io/machine-learning/svd-image-compression/