About this document
BinHD: Efficient Binary Hyperdimensional Computing by S D is a document available to read on EtoBox.
The document presents BinHD, a novel binary learning framework for Hyperdimensional (HD) computing that enables efficient training and inference using binary hypervectors. BinHD reduces computational costs and memory requirements while maintaining classification accuracy by utilizing a hardware-friendly Hamming distance metric instead of the costly Cosine similarity. Evaluations demonstrate that BinHD achieves significant energy efficiency and speed improvements over existing HD computing algorithms across
- Author
- S D
- Language
- EN