Norm

•What is the appropriate and differentiable norm when given data points with outliers? (median is not differentiable)
•How do you define outliers? What is the probabilistic meaning of outliers?

첫 번째 질문의 답이 p-norm with high p였던 거 같아서 p-norm, 2-norm, frobenius norm을 찾아보다가 svd, eigenvalue decomposition을 찾아보다 왜 p-norm with high p인 지 아직도 이해를 못함.

일단 induced norm 인가 schatten norm인 지 모르겠음.

https://chrischoy.github.io/research/matrix-norms/#:~:text=In%20another%20word%2C%20matrix%20p,for%20a%20unit%20vector%20e.&text=Largest%20singular%20value%20of%20a,not%20necessarily%20the%20largest%20norm.

Interesting Properties of Matrix Norms and Singular Values

https://math.stackexchange.com/questions/33083/what-is-the-difference-between-the-frobenius-norm-and-the-2-norm-of-a-matrix#:~:text=The%202%2Dnorm%20(spectral%20norm,between%20all%20the%20singular%20values.

What is the difference between the Frobenius norm and the 2-norm of a matrix?

일단 svd를 철저히 이해하면 어떻게든 되려나

https://math.stackexchange.com/questions/320220/intuitively-what-is-the-difference-between-eigendecomposition-and-singular-valu

Intuitively, what is the difference between Eigendecomposition and Singular Value Decomposition?

SVD in Machine Learning: Underdetermined Least Squares

https://towardsdatascience.com/underdetermined-least-squares-feea1ac16a9

Steve Brunton

https://www.youtube.com/watch?v=PjeOmOz9jSY

PCA and rigid transformation

https://towardsdatascience.com/principal-component-analysis-pca-8133b02f11bd

이 포스트 나중에 정리해야지. 일단 덤프