Computing with High-Dimensional Vectors

  • Authors:
    Pentti Kanerva (UC/Berkeley)
    Publication ID:
    P093136
    Publication Type:
    Paper
    Received Date:
    15-Feb-2018
    Last Edit Date:
    19-Feb-2018
    Research:
    2702.001 (University of California/Berkeley)

Abstract

"Computing power'' usually implies numbers, speed, high-precision arithmetic, large memory, operating with uncompromising reliability and reasonable demands on energy. Contradiction in these requirements is leading to leveling off of the amount of computing per dollar, referred to popularly as "Moore's Law running out of steam.'' Here we consider the possibility of increasing computing power by computing with high-dimensional (10,000-D) vectors that need not be ultra-reliable nor precise. The theory can guide the development of nanotechnology, and applications abound in areas now the domain of AI, neural nets, and deep learning.

4819 Emperor Blvd, Suite 300 Durham, NC 27703 Voice: (919) 941-9400 Fax: (919) 941-9450

Important Information for the SRC website. This site uses cookies to store information on your computer. By continuing to use our site, you consent to our cookies. If you are not happy with the use of these cookies, please review our Cookie Policy to learn how they can be disabled. By disabling cookies, some features of the site will not work.