Automated optimized parameters for T-distributed stochastic neighbor embedding improve visualization and analysis of large datasets

By Anna C. Belkina, Chris Ciccolella, R. Anno, Richard L. Halpert, Josef Spidlen, J. Snyder-Cappione. 2019

553 citations recorded by the source corpus.

Accurate and comprehensive extraction of information from high-dimensional single cell datasets necessitates faithful visualizations to assess biological populations. A state-of-the-art algorithm for non-linear dimension reduction, t-SNE, requires multiple heuristics and fails to produce clear representations of datasets when millions of cells are projected. We develop opt-SNE, an automated toolkit for t-SNE parameter selection that utilizes Kullback-Leibler divergence evaluation in real time to tailor the early exaggeration and overall number of gradient descent iterations in a dataset-specific manner. The precise calibration of early exaggeration together with opt-SNE adjustment of gradient descent learning rate dramatically improves computation time and enables high-quality visualization of large cytometry and transcriptomics datasets, overcoming limitations of analysis tools with hard-coded parameters that often produce poorly resolved or misleading maps of fluorescent and mass cytometry data. In summary, opt-SNE enables superior data resolution in t-SNE space and thereby more accurate data interpretation. Visualisation tools that use dimensionality reduction, such as t-SNE, provide poor visualisation on large data sets of millions of observations. Here the authors present opt-SNE, that automatically finds data set-tailored parameters for t-SNE to optimise visualisation and improve analysis.

Subfields: Biophysics, Molecular Biology, Representation & Self-Supervised Learning

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Biophysics

In Molecular Biology

In Representation & Self-Supervised Learning

Citations in the corpus

Interpretable Machine Learning: Fundamental Principles and 10 Grand Challenges (2021, background): “Multiple research (for example (Belkina et al., 2019; Wang et al., 2020b)) aimed to solve this problem for the most influential hyperparameters, but the problem still exists, and the set of hyperparameters remains data-dependent.”
Understanding How Dimension Reduction Tools Work: An Empirical Approach to Deciphering t-SNE, UMAP, TriMAP, and PaCMAP for Data Visualization (2020, methodology): “…methods that focus on preserving local structure are t-SNE (van der Maaten and Hinton, 2008), LargeVis (Tang et al., 2016), and UMAP (McInnes et al., 2018) (see van der Maaten et al., 2009; Bellet et al., 2013; Belkina et al., 2019, for a survey of these methods and how to tune their parameters).”
openTSNE: a modular Python library for t-SNE dimensionality reduction and embedding (2019, methodology): “We adopt the learning rate η = N/ρ proposed by Belkina et al. (2019), who suggest using a global learning rate computed from the early exaggeration factor ρ .”
A Unifying Perspective on Neighbor Embeddings along the Attraction-Repulsion Spectrum (2020, unknown): “The learning rate was set to η = n / max( ρ, ρ early ) (Belkina et al., 2019).”
Embedding to reference t-SNE space addresses batch effects in single-cell classification (2019, methodology): “When using t-SNE on larger data sets, the standard learning rate = 200 has been shown to lead to slower convergence and requires more iterations to achieve consistent embeddings (Belkina et al., 2019).”
NOMAD Projection (2025, unknown): “We set our initial learning rate to n/ 10 , a factor of 10 less than the t-SNE convention presented in Belkina et al. [2], as we find it helps with convergence.”