Classifying Handwritten Digits Using Machine Learning
The MNIST data set is a data set that contains 60,000 small squared pixel grayscale images.
The goal of this project was to be able to use some unsupervised learning techniques from the first half of the class in order to identify these digits. From class, we learned that unsupervised machine learning consists of using data with no labels to create a class as an output. For the project we use three different algorithms in attempts to classify the digit images: Single Multivariate Bernoulli Distribution, Kmeans Clustering, and a Mixture of Multivariate Bernoulli Models
Click the link above the photo to see our approach in writing and code.
