An SVD based classifier for MNIST digits The attached notebook is research into building a classifier for the MNIST data set using least square distance to the orthonormal basis corresponding to each digit Please note that this notebook is under revision and needs additional commentary to fully contextualize the data.
The one comment that I would like to makesure is clear is that the first graph describing percentage of missed classifications is improperly labeled. the y axis is already in percentage form, however it reads as if it has not yet been converted.
Work completed during the Spring of 2022 when working on computational linear algebra and computer vision topics