When asked about the number of dimensions that an image has, photographers, painters, illustrators, and almost everyone else on this planet will agree that images are two-dimensional objects. Only machine learning practitioners see images differently. For us, every pixel in a black and white image is a separate dimension. Dimensions expand even more with colored images, but that's something for later. We see each pixel as a separate dimension so that we can deal with each pixel and its value as a unique feature that defines the image, along with the other pixels (features). So, unlikeTino Sehgal, we can sometimes end up working with 4,000 dimensions.
The ModifiedNational Instituteof Standardsand Technology(MNIST) dataset is a collection of handwritten digits that is commonly used inimage processing. Due to its popularity, it...
