Quantum-classical truncated Newton method for high-dimensional energy landscapes

TL;DR

This paper introduces a quantum-classical hybrid optimization algorithm tailored for high-dimensional energy landscapes, enhancing neural network training and deep generative model parameter tuning with potential polynomial or exponential speedups.

Contribution

It presents a novel truncated saddle-free Newton's method combined with quantum subroutines, offering a new hybrid approach for efficient high-dimensional optimization.

Findings

Algorithm achieves at least as good performance as classical methods.

Potential for polynomial speedup in optimization tasks.

Omission of classical read-out could lead to exponential speedup.

Abstract

We develop a quantum-classical hybrid algorithm for function optimization that is particularly useful in the training of neural networks since it makes use of particular aspects of high-dimensional energy landscapes. Due to a recent formulation of semi-supervised learning as an optimization problem, the algorithm can further be used to find the optimal model parameters for deep generative models. In particular, we present a truncated saddle-free Newton's method based on recent insight from optimization, analysis of deep neural networks and random matrix theory. By combining these with the specific quantum subroutines we are able to exhaust quantum computing in order to arrive at a new quantum-classical hybrid algorithm design. Our algorithm is expected to perform at least as well as existing classical algorithms while achieving a polynomial speedup. The speedup is limited by the required classical read-out. Omitting this requirement can in theory lead to an exponential speedup.