Vector Symbolic Architectures as a Computing Framework for Emerging Hardware

Area of Science:

• Computer Science
• Artificial Intelligence
• Cognitive Science

Background:

• Emerging hardware necessitates novel computing frameworks.
• Artificial Intelligence (AI) requires efficient cognitive operation processing.
• Conventional computing faces challenges with complex combinatorial search problems.

Purpose of the Study:

• To review recent advancements in Vector Symbolic Architectures (VSA).
• To highlight VSA's suitability for AI and stochastic, emerging hardware.
• To demonstrate VSA's potential as a computational framework and abstraction layer.

Main Methods:

• Review of Vector Symbolic Architectures (VSA) principles and operations.
• Analysis of VSA's algebraic structure for high-dimensional vector manipulation.
• Illustration of VSA's "computing in superposition" feature.

Main Results:

• VSA provides simple, powerful operations on high-dimensional vectors supporting modern computing needs.
• VSA's "computing in superposition" enables efficient solutions for AI's combinatorial search problems.
• VSA demonstrates potential for computational universality and distributed representations.

Conclusions:

• Vector Symbolic Architectures (VSA) are a promising framework for AI and emerging hardware.
• VSA offers a unique approach to computing with distributed representations.
• VSA can serve as an abstraction layer for novel computing architectures like neuromorphic computing.