The Concept of ‘Cat Face’
Paul Taylor: Machine Learning, 11 August 2016
“... images: fifty of them are of the letter ‘i’, each in different handwriting in shades of grey on a white background; the other fifty, similarly presented, are of ‘j’s. If each image is 32 pixels high and 32 pixels wide, it can be represented as a single point in a 1024-dimensional space (322 = 1024), where each dimension corresponds to a ... ”