Summary: Hyperdimensional computing, a promising alternative to traditional computing

Hyperdimensional computing is a new computing paradigm that uses vectors to represent concepts, and hyperdimensional algebra to manipulate them. The system was developed in the mid-1990s by Pentti Kanerva and Tony Plate, and demonstrates that concepts can be manipulated symbolically through mathematical operations. Hyperdimensional computing is transparent and offers resilience to computing errors. It also allows for the creation of low-power hardware computing that performs data storage and processing on the same hardware. Although still in its early stages, hyperdimensional computing has significant potential and offers an alternative to traditional computing. Researchers are working toward hybrid systems that use neural networks to map items in the physical world to hypervectors, with hyperdimensional algebra taking over.

Related articles

Quanta Magazine

By imbuing enormous vectors with semantic meaning, we can get machines to reason more abstractly — and efficiently — than before.

Read the complete article at: www.quantamagazine.org