Saved in:
| Main Authors: | , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2602.00116 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866915764100923392 |
|---|---|
| author | Dejonghe, Hanne Leroux, Sam |
| author_facet | Dejonghe, Hanne Leroux, Sam |
| contents | Hyperdimensional computing (HDC) offers lightweight learning for energy-constrained devices by encoding data into high-dimensional vectors. However, its reliance on ultra-high dimensionality and static, randomly initialized hypervectors limits memory efficiency and learning capacity. Therefore, we propose Trainable Hyperdimensional Computing (THDC), which enables end-to-end HDC via backpropagation. THDC replaces randomly initialized vectors with trainable embeddings and introduces a one-layer binary neural network to optimize class representations. Evaluated on MNIST, Fashion-MNIST and CIFAR-10, THDC achieves equal or better accuracy than state-of-the-art HDC, with dimensionality reduced from 10.000 to 64. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2602_00116 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | THDC: Training Hyperdimensional Computing Models with Backpropagation Dejonghe, Hanne Leroux, Sam Machine Learning Hyperdimensional computing (HDC) offers lightweight learning for energy-constrained devices by encoding data into high-dimensional vectors. However, its reliance on ultra-high dimensionality and static, randomly initialized hypervectors limits memory efficiency and learning capacity. Therefore, we propose Trainable Hyperdimensional Computing (THDC), which enables end-to-end HDC via backpropagation. THDC replaces randomly initialized vectors with trainable embeddings and introduces a one-layer binary neural network to optimize class representations. Evaluated on MNIST, Fashion-MNIST and CIFAR-10, THDC achieves equal or better accuracy than state-of-the-art HDC, with dimensionality reduced from 10.000 to 64. |
| title | THDC: Training Hyperdimensional Computing Models with Backpropagation |
| topic | Machine Learning |
| url | https://arxiv.org/abs/2602.00116 |