Quantum sequel of neural network training

TL;DR

This paper demonstrates that quantum annealing platforms can efficiently train classical neural networks, outperforming traditional methods in scaling, and offers a physics-based perspective on the training process as a dynamical phase transition.

Contribution

It provides the first experimental evidence that quantum annealing can be used for neural network training, showing superior scaling and potential for deep networks.

Findings

Quantum annealing achieves better scaling than classical backpropagation.

Training via quantum annealing can be enhanced with a fully coherent quantum platform.

Modest quantum annealers can effectively train deep neural networks layer-wise.

Abstract

Training of neural networks (NNs) has emerged as a major consumer of both computational and energy resources. Quantum computers were coined as a root to facilitate training, but no experimental evidence has been presented so far. Here we demonstrate that quantum annealing platforms, such as D-Wave, can enable fast and efficient training of classical NNs, which are then deployable on conventional hardware. From a physics perspective, NN training can be viewed as a dynamical phase transition: the system evolves from an initial spin glass state to a highly ordered, trained state. This process involves eliminating numerous undesired minima in its energy landscape. The advantage of annealing devices is their ability to rapidly find multiple deep states. We found that this quantum training achieves superior performance scaling compared to classical backpropagation methods, with a clearly higher scaling exponent (1.01 vs. 0.78). It may be further increased up to a factor of 2 with a fully coherent quantum platform using a variant of the Grover algorithm. Furthermore, we argue that even a modestly sized annealer can be beneficial to train a deep NN by being applied sequentially to a few layers at a time.