DARPA-Funded Research: Applying Memcomputing to AI

The Di Ventra Group at UC San Diego has been awarded $500,000 in funding from the Defense Advanced Research Projects Agency (DARPA). The DARPA funds will be used to research the application of a physics-based approach (memcomputing) to artificial intelligence. Our current focus is applying memcomputing to the unsupervised learning phase of training Deep Belief Networks. More details can be found in the UCSD press release: Riding the Third Wave of AI without Quantum Computing.

Sketch of a Memcomputing Architecture

Apart from the input/output and a control unit, which directs the machine on what problem to solve, all computation is done by a memory unit, a “computational memory.” From F.L. Traversa and M. Di Ventra, IEEE Trans. Neural Networks Learn. Sys. 26, 2702 (2015). © 2015 IEEE.

Exponential vs. Polynomial Scaling

Demonstration that a memcomputing solver (named Falcon in the figure) outperforms, by orders of magnitude, state-of-the-art algorithms in solving difficult computational problems. From F. Sheldon, P. Cicotti, F.L. Traversa and M. Di Ventra, IEEE Trans. Neural Networks Learn. Sys. (2019). © 2019 IEEE.