Hands-On Image Processing with Python

Bayes classifier (Gaussian generative model)

As we have seen in the last section, the 1-NN classifier yielded a 3.09% test error rate on the MNIST data set of handwritten digits. In this section, we will build a Gaussian generative model that does almost as well, while being significantly faster and more compact. Again, we need to load the MNIST training and test dataset first, as we did last time. Next, let's fit a Gaussian generative model to the training dataset.