Capacity Analysis of Vector Symbolic Architectures
Offered By: Simons Institute via YouTube
Course Description
Overview
Explore a comprehensive lecture on the capacity analysis of Vector Symbolic Architectures (VSAs) in hyperdimensional computing. Delve into the theoretical aspects of VSAs, focusing on their representation capacities and the dimensions required for various symbolic tasks. Examine four common VSAs: MAP-I, MAP-B, and two based on sparse binary vectors. Investigate the performance of these architectures in set membership testing and estimating set intersection sizes. Learn about a novel Hopfield network variant and its capabilities in VSA-typical tasks. Discover the connections between VSAs, sketching algorithms, and Bloom filters. Gain insights from this collaborative research presented by Ken Clarkson from IBM Research, offering new bounds on VSA capacities and expanding the theoretical understanding of hyperdimensional computing frameworks.
Syllabus
Capacity Analysis of Vector Symbolic Architectures
Taught by
Simons Institute
Related Courses
On Quantum Linear Algebra for Machine Learning - Quantum ColloquiumSimons Institute via YouTube DataSketches: A Production Quality Sketching Library for Big Data Analysis
Databricks via YouTube On Quantum Linear Algebra for Machine Learning - IPAM at UCLA
Institute for Pure & Applied Mathematics (IPAM) via YouTube Streaming and Learning Algorithms - Session 7C
IEEE via YouTube Sketching Algorithms for Max-DICUT and Other CSPs
Simons Institute via YouTube