Distributed stochastic neighbor embedding

The stochastic neighbor embedding (SNE) problem is formulated as follows: we have a dataset with points described by a multidimensional variable with a dimension of space substantially higher than three. It is necessary to obtain a new variable that exists in a two-dimensional or three-dimensional space that would maximally preserve the structure and patterns in the original data. The difference between t-SNE and the classic SNE lies in the modifications that simplify the process of finding the global minima. The main modification is replacing the normal distribution with the Student's t-distribution for low-dimensional data. SNE begins by converting the multidimensional Euclidean distance between ...

Get Hands-On Machine Learning with C++ now with the O’Reilly learning platform.

O’Reilly members experience books, live events, courses curated by job role, and more from O’Reilly and nearly 200 top publishers.