Approximated and User Steerable tSNE for Progressive Visual Analytics

By Nicola Pezzotti, Boudewijn P. F. Lelieveldt, L. V. D. van der Maaten, Thomas Höllt, Elmar Eisemann, Anna Vilanova. 2015

303 citations recorded by the source corpus.

Progressive Visual Analytics aims at improving the interactivity in existing analytics techniques by means of visualization as well as interaction with intermediate results. One key method for data analysis is dimensionality reduction, for example, to produce 2D embeddings that can be visualized and analyzed efficiently. t-Distributed Stochastic Neighbor Embedding (tSNE) is a well-suited technique for the visualization of high-dimensional data. tSNE can create meaningful intermediate results but suffers from a slow initialization that constrains its application in Progressive Visual Analytics. We introduce a controllable tSNE approximation (A-tSNE), which trades off speed and accuracy, to enable interactive data exploration. We offer real-time visualization techniques, including a density-based solution and a Magic Lens to inspect the degree of approximation. With this feedback, the user can decide on local refinements and steer the approximation level during the analysis. We demonstrate our technique with several datasets, in a real-world research scenario and for the real-time analysis of high-dimensional streams to illustrate its effectiveness for interactive data analysis.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition

How researchers see this paper

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

In Artificial Intelligence

In Computer Vision and Pattern Recognition

Citations in the corpus

Automated optimized parameters for T-distributed stochastic neighbor embedding improve visualization and analysis of large datasets (2019, methodology): “A promising approach that may be integrated with opt-SNE is the smart EE adjustment implemented in A-tSNE (approximated t-SNE) 35 algorithm where EE is removed gradually and on a per-point basis.”
Hierarchical Stochastic Neighbor Embedding (2016, unknown): “Using approximations it can be reduced to O ( n log ( n )) [PL ∗ 15, VDM14].”
SOM-based embedding improves efficiency of high-dimensional cytometry data analysis (2019, methodology): “For example, the widely used tSNE [24] has formed the basis 30 for faster ASNE [17] and HSNE [16], and was further improved in OptSNE and FItSNE [3, 14] and accelerated using GPU by Chan et al.”
Visual analysis of mass cytometry data by hierarchical stochastic neighbour embedding reveals rare cell types (2017, methodology): “Similarities between cells on the data level are de ﬁ ned by building an approximated, weighted k-nearest neighbor (kNN) graph 16 using the Euclidian distances based on the complete marker expression (Fig.”
T-SNE-CUDA: GPU-Accelerated T-SNE and its Applications to Modern Data (2018, methodology): “To solve the approximate nearest neighbor problem, Pezzotti et al. [19] use a random forest of approximate K-d trees, which allowed them to compute the nearest neighbors in a faster manner, however their approach still suffers from high dimen-sion.”
Towards better analysis of machine learning models: A visual analytics perspective (2017, background): “priate views or next steps in the iterative and progressive analysis process. From the information visualization community, researchers have designed a number of techniquesforuser initiative guidance [14, 30, 33, 34, 49], where the user is the active participant in improving and rening the performan”
Linear tSNE optimization for the Web (2018, methodology): “[20] observed that the computation of the k-nearest neighborhood graph for high-dimensional spaces using the Vantage-Point Tree is affected by the curse of dimensionality, limiting the efficiency of the computation.”