Detail View

Multi-Centroid Hyperdimensional Computing for Compact IMC Arrays via Dimension Pruning

Citations

WEB OF SCIENCE

Citations

SCOPUS

Metadata Downloads

DC Field Value Language
dc.contributor.author Kang, Do Yeong -
dc.contributor.author Oh, Yeong Hwan -
dc.contributor.author Hwang, Chanwook -
dc.contributor.author Kim, Jinhee -
dc.contributor.author Jeon, Kang Eun -
dc.contributor.author Ko, Jong Hwan -
dc.date.accessioned 2026-11-17T16:10:11Z -
dc.date.available 2026-11-17T16:10:11Z -
dc.date.created 2026-09-30 -
dc.date.issued 2026-11 -
dc.identifier.issn 2372-2541 -
dc.identifier.uri https://scholar.dgist.ac.kr/handle/20.500.11750/61181 -
dc.description.abstract The implementation of Hyperdimensional Computing (HDC) on In-Memory Computing (IMC) architectures faces fundamental challenges due to dimensional mismatches between high-dimensional vectors and IMC array constraints. Traditional HDC approaches utilize vectors that exceed IMC array dimensions, causing computational resource underutilization and leaving most columns unused during similarity calculations. Existing partitioning methods fail to reduce computation cycles, with challenges intensified for compact IMC arrays (64×64 and 128×128). This paper presents MEMHD, a Memory-Efficient Multi-centroid HDC framework that replaces traditional single-vector-per-class with multiple class vectors to achieve full IMC array utilization and enable single-cycle inference through clustering-based initialization and quantization-aware training. To address compact IMC array deployment, we also propose DiP-MEMHD (Dimension Pruned MEMHD), which prunes dimensions while preserving critical information through calibration and variance-based weighting. As a result, MEMHD achieves 10.60× better memory efficiency at comparable accuracy and 14.61% higher accuracy with the same memory usage compared to existing binary HDC baselines. DiP-MEMHD introduces multi-centroid optimized pruning metric, showing 2.47% higher accuracy than existing pruning metrics and achieving 22.4% higher accuracy than state-of-the-art binary HDC models on compact IMC arrays with 6-8× better simulated energy efficiency. These properties make the framework suitable for always-on IoT classification on compact edge devices, such as wearable activity recognition and other sensor-based tasks under tight latency and power budgets. To the best of our knowledge, this is the first multi-centroid HDC framework specifically optimized for IMC array constraints, enabling practical deployment for real-time IoT applications. -
dc.language English -
dc.publisher Institute of Electrical and Electronics Engineers Inc. -
dc.title Multi-Centroid Hyperdimensional Computing for Compact IMC Arrays via Dimension Pruning -
dc.type Article -
dc.identifier.doi 10.1109/JIOT.2026.3730047 -
dc.identifier.wosid 001916151400008 -
dc.identifier.scopusid 2-s2.0-105050229536 -
dc.identifier.bibliographicCitation IEEE Internet of Things Journal, v.13, no.21, pp.48701 - 48718 -
dc.description.isOpenAccess FALSE -
dc.subject.keywordAuthor Arrays -
dc.subject.keywordAuthor Accuracy -
dc.subject.keywordAuthor Modeling -
dc.subject.keywordAuthor Vectors -
dc.subject.keywordAuthor In-memory computing -
dc.subject.keywordAuthor Training -
dc.subject.keywordAuthor Measurement -
dc.subject.keywordAuthor Memory -
dc.subject.keywordAuthor Internet of Things -
dc.subject.keywordAuthor Learning (artificial intelligence) -
dc.subject.keywordAuthor Dimension pruning -
dc.subject.keywordAuthor edge computing -
dc.subject.keywordAuthor hyperdimensional computing (HDC) -
dc.subject.keywordAuthor in-memory computing (IMC) -
dc.subject.keywordAuthor Internet of Things (IoT) -
dc.subject.keywordAuthor multicentroid associative memory (AM) -
dc.citation.endPage 48718 -
dc.citation.number 21 -
dc.citation.startPage 48701 -
dc.citation.title IEEE Internet of Things Journal -
dc.citation.volume 13 -
dc.description.journalRegisteredClass scie -
dc.description.journalRegisteredClass scopus -
dc.relation.journalResearchArea Computer Science;Engineering;Telecommunications -
dc.relation.journalWebOfScienceCategory Computer Science, Information Systems;Engineering, Electrical & Electronic;Telecommunications -
dc.type.docType Article -
Show Simple Item Record

File Downloads

  • There are no files associated with this item.

공유

qrcode
공유하기

Related Researcher

전강은
Jeon, Kang Eun전강은

Department of Electrical Engineering and Computer Science

read more

Total Views & Downloads

???jsp.display-item.statistics.view???: , ???jsp.display-item.statistics.download???: