GPUMAP: GPU-Accelerated Progressive Uniform Manifold Approximation and Projection
Organizations: Sungkyunkwan University
Abstract
Dimensionality reduction (DR) techniques are fundamental to the analysis of high-dimensional data, yet their substantial computational cost limits their use in interactive settings. To address this limitation, we present GPU-accelerated Progressive Uniform Manifold Approximation and Projection (GPUMAP), an efficient progressive DR algorithm designed for interactive use. We identify three atomic operations within UMAP and restructure them into a progressive pipeline. As part of this adaptation, we introduce modifications to UMAP, including operation queues, instance-wise local join, and scheduling strategies. We further accelerate computation by leveraging GPU parallelism, enabling interactive analysis of million-scale datasets on commodity GPU hardware. Finally, we employ our implementation to systematically study the behavior of progressive DR, investigating how projections evolve over time relative to the ground-truth projection and how biased data streams influence the quality and stability of the resulting projections.
Figures & tables
| Dataset | Size ( dims) | Data Type |
| EMNIST Digits [ 4 ] | 0, 280,000 784 | Image |
| Covertype [ 3 ] | 0, 581,012 54 | Table |
| SVHN [ 19 ] | 0, 630,420 768 | Image |
| SIFT1M [ 11 ] | 1,000,000 128 | Image |
| MARC [ 13 ] | 1,200,000 768 | Text |
| Mouse Brain [ 1 ] | 1,306,127 50 | Genomic |
| Dataset | Time (s) | Final Recall | Norm. Fidelity |
|---|---|---|---|
| EMNIST Digits | |||
| Covertype | |||
| SVHN | |||
| SIFT1M | |||
| MARC | |||
| Mouse Brain |
Dimensionality Reduction
Dimensionality reduction maps high-dimensional data to compact representations for visualization, prediction, compression, and scientific modeling.
See all 159 papers