Transformation using SVD |
|
|
is called SVD transform matrix
|
- Essentially,
is just a rotation of ![](images/img617.png)
|
- Dimensionality of
is ![](images/img6.png)
|
different basis vectors than the original space
|
- Columns of
give the basis vectors in rotated space
|
shows how each dimension can be represented as a linear combination of other dimensions
|
|
- Columns are input basis vectors
|
shows how each object can be represented as a linear combination of other objects
|
|
- Columns are output basis vectors
|
- Lengths of vectors are preserved
|
|