t-SNE
T-distributed Stochastic Neighbor Embedding is a non-linear dimensionality reduction algorithm designed explicitly to visualize high-dimensional data in 2D.
Think of It Like This
Like squishing a complex 3D galaxy of stars onto a flat map so that stars that were close together remain tightly grouped on the paper.
t-SNE calculates the probability that two points are neighbors in high-dimensional space and attempts to perfectly replicate that probability distribution in 2D space. It is incredible for visually identifying clusters in complex datasets (like visualizing word embeddings). However, it is computationally expensive and does not preserve global distances.