t-SNE (t-distributed stochastic neighbour embedding) is a method used in statistics to visualise higher-dimensional data in a 2D or 3D map. For classification problems, this can be quite useful for understanding what’s going on.
Parallelisation
t-SNE has a time complexity of . Thus, a parallelised implementation is important. Instead of using scikit-learn (which runs on a single thread), we should use the tsnecuda package.
https://www.dailydoseofds.com/formulating-and-implementing-the-t-sne-algorithm-from-scratch/