When enabled, t-Distributed Stochastic Neighbor Embedding (t-SNE) is used to visualize groupings resulting from dimensionality reduction. Java implementation by Dr. Leif Jonsson, Ph.D. at Ericsson Machine Learning and Analytics Platforms uses parallelized Barnes Hut approximation to vastly improve computational performance. The resulting x-y coordinates are embedded in the FCS file for every cell within a region.

Van Der Maaten, L. (2014). Accelerating t-SNE using tree-based algorithms. The Journal of Machine Learning Research, 15(1), 3221-3245.

GitHub Link

Last updated