High-Performance Hyperdimensional Computing in JAX
Open-source JAX library for hyperdimensional computing and vector symbolic architectures. Publication pending, 2026.
What It Does
A single functional API for four VSA models — BSC, MAP, HRR, and Fourier HRR — that works with every JAX transform: jit, vmap, pmap, grad.
Results
- 1–3x faster than NumPy on CPU; outperforms PyTorch baselines on GPU
- XLA compilation fuses kernels and scales across CPUs, GPUs, and TPUs
- Benchmarked on EU language identification, EMG gesture recognition, and VoiceHD
Features
- Random and feature encoders, centroid classifiers, gradient-based learning
- Built for large-scale HDC research in ML, neuro-symbolic AI, and edge computing
ArXiv link coming soon.