Summary

In this chapter, we were introduced to t-Distributed Stochastic Neighbor Embeddings as a means of visualizing high-dimensional information that may have been produced from prior processes such as PCA or autoencoders. We discussed the means by which t-SNEs produce this representation and generated a number of them using the MNIST and Wine datasets and scikit-learn. In this chapter, we were able to see some of the power of unsupervised learning because PCA and t-SNE were able to cluster the classes of each image without knowing the ground truth result. In the next chapter, we will build on this practical experience as we look into the applications of unsupervised learning, including basket analysis and topic modeling.