Accelerating t-SNE using tree-based algorithms
Laurens van der Maaten · 2014
The paper investigates the acceleration of t-SNE—an embedding technique that is com-monly used for the visualization of high-dimensional data in scatter plots—using two tree-based algorithms. In particular, the paper develops variants of the Barnes-Hut algorithm and of the dual-tree algorithm that approximate the gradient used for learning t-SNE em-beddings in O(N logN). Our experiments show that the resulting algorithms substantially accelerate t-SNE, and that they make it possible to learn embeddings of data sets with millions of objects. Somewhat counterintuitively, the Barnes-Hut variant of t-SNE appears to outperform the dual-tree variant.
Cited by 2,226 works. doi:10.5555/2627435.2697068
Authors
References
- Learning Multiple Layers of Features from Tiny Images
- Improving neural networks by preventing co-adaptation of feature detectors
- Nonlinear Dimensionality Reduction by Locally Linear Embedding
- Visualizing Data using t-SNE
- A Global Geometric Framework for Nonlinear Dimensionality Reduction
- Learning Phrase Representations using RNN Encoder–Decoder for Statistical Machine Translation
- Learning methods for generic object recognition with invariance to pose and lighting
- An Algorithm for Finding Best Matches in Logarithmic Expected Time
- Caffe: Convolutional Architecture for Fast Feature Embedding
- Scalable Recognition with a Vocabulary Tree