Reproduce “Laplace-HDC: Understanding the geometry of binary hyperdimensional computing” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Saeid Pourmand, Wyatt D. Whiting, Alireza Aghasi, Nicholas F. Marshall

Laplace-HDC combines binary hyperdimensional computing with a Laplace-like similarity geometry, whose expected encoded similarity follows a product of feature similarities. For images, the authors add local Haar-filter features before encoding. On Fashion-MNIST, Haar-feature Laplace-HDC with 1D-Cyclic permutations and Binary SGD reaches 86.65 ± 0.36% accuracy, versus 81.19 ± 1.81% for RFF-HDC; this compares complete configurations, not kernels alone.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.