ADVANCED HYPERDIMENSIONAL MATHEMATICS FOR ADAPTIVE INFORMATION PROCESSING

Abstract

HyperDimensional Computing (HDC) is introduced as an alternative paradigm that mimics important brain functionalities towards high-efficiency and noise-tolerant computation. There are limitations in using current HDC mathematics to encode and represent complex data structures or enable reasoning. In this proposal, we aim to develop flexible HDC mathematics that can effectively encode and represent complex data structures in high-dimensional space. Our mathematics defines attention in hyperspace, expands HDC functionality to stochastic computing, and enables correlative information association. We accordingly develop encoding methods that exploit neural dynamics to generate self-recovery representation with natural adaptation to changes in data and environment. Finally, we develop solutions supporting brain-like learning, cognitive computing, reasoning by exploiting high-dimensional information extraction and data reconstruction.

Document Details

Document Type
DoD Grant Award
Publication Date
Mar 07, 2023
Source ID
FA95502210253

Entities

People

  • Mohsen Imani

Organizations

  • Air Force Office of Scientific Research
  • United States Air Force
  • University of California, Irvine

Tags

Fields of Study

  • Computer science

Readers

  • Distributed Systems and Data Platform Development
  • Neural Network Machine Learning.
  • Theoretical Analysis.

Technology Areas

  • AI & ML
  • AI & ML - Neural Networks
  • Space