Program Summary
Machine learning and artificial intelligence techniques are currently being applied in a diverse number of fields, including molecular simulation, many-body physics, classification, and computational optimization. However, progress in addressing these types of problems is being slowed or stopped when the problem complexity grows exponentially with problem size. Moreover, even when these complexity barriers are overcome, the impact of machine learning solutions are often mitigated by the high energy cost of training and operating the machine learning systems.
In principle, both of these fundamental obstacles – exponentially growing complexity and energy inefficiency – might be overcome using high-coherence quantum annealers, which are a specific type of quantum computing technology.
DARPA’s Reversible Quantum Machine Learning and Simulation (RQMLS) AIE opportunity aims to: (1) explore the fundamental limits of reversible quantum annealers; (2) quantitatively predict the computational utility of these systems for machine learning tasks in simulation, many-body physics, classification, optimization, and other fields; and (3) design experimental tests for these predictions that can be carried out on small-scale systems. If successful, these small-scale systems could be scaled to much larger, potentially transformative systems.