AIsing: Analog CMOS Ising Machines for Natural Computing

Conventional von Neumann computing architectures are unable to efficiently solve many of today’s practically relevant combinatorial optimization problems, e.g. vehicle routing, airplanes scheduling, work-force scheduling, network diagnosis, training artificial intelligence.

Ising Machine technology, inspired by nature, is an attractive non-conventional computing alternative for solving complex combinatorial optimization problems.The objective of this proposal is to address fundamental and applied challenges at the frontiers of Ising Machine technology for large spin networks and propose solutions to accomplish their difficult requirements in terms of energy efficiency, speed, accuracy, carbon footprint and cost.

In this project, we aim to explore the excellent opportunities that analog CMOS implementation can offer to natural computing technology and develop a proof-of-concept energy-efficient analog CMOS Ising Machine able to outperform state-of-the-art large spin networks in a wide range of complex combinatorial optimization problems.This project will outline the foundation for a new research area with huge impact on various emerging technologies and applications, e.g. artificial intelligence, machine learning, high-speed data encryption and next generation communication.

With this project, we hope to be able to give Sweden a leading position in the development of a key next-generation computing technology.