Interpretable Quantum Advantage in Neural Sequence Learning

Abstract

Quantum neural networks have been widely studied in recent years, given their potential practical utility and recent results regarding their ability to efficiently express certain classical data. However, analytic results to date rely on assumptions and arguments from complexity theory. Due to this, there is little intuition as to the source of the expressive power of quantum neural networks or for which classes of classical data any advantage can be reasonably expected to hold. Here, we study the relative expressive power between a broad class of neural network sequence models and a class of recurrent models based on Gaussian operations with non-Gaussian measurements. We explicitly show that quantum contextuality is the source of an unconditional memory separation in the expressivity of the two model classes. Additionally, as we are able to pinpoint quantum contextuality as the source of this separation, we use this intuition to study the relative performance of our introduced model on a standard translation data set exhibiting linguistic contextuality. In doing so, we demonstrate that our introduced quantum models are able to outperform state of the art classical models even in practice.

30 pages, 9 figures

References in corpus (8)

• Supplementary information for "Quantum supremacy using a programmable superconducting processor"
• Quantum computational advantage using photons
• An introduction to quantum machine learning
• Strong quantum computational advantage using a superconducting quantum processor
• Phase-Programmable Gaussian Boson Sampling Using Stimulated Squeezed Light
• Beyond Barren Plateaus: Quantum Variational Algorithms Are Swamped With Traps
• Graphical calculus for Gaussian pure states
• State-independent quantum contextuality for continuous variables

Cited by in corpus (17)

• Does provable absence of barren plateaus imply classical simulability?
• Trainability barriers and opportunities in quantum generative modeling
• Efficient classical algorithms for simulating symmetric quantum systems
• Quantum neural networks form Gaussian processes
• Entanglement-induced provable and robust quantum learning advantages
• Digital-analog quantum learning on Rydberg atom arrays
• Quantum Adjoint Convolutional Layers for Effective Data Representation
• Cost of Locally Approximating High-Dimensional Ground States of Contextual Quantum Models
• Harvesting Contextuality from the Vacuum
• Unconditionally separating noisy ${\mathsf{QNC}}^0$ from bounded polynomial threshold circuits of constant depth
• Contextual quantum metrology
• Harnessing Quantum Dynamics for Robust and Scalable Quantum Extreme Learning Machines
• Demonstration of sequential processors with quantum advantage and analysis of classical performance limits
• k-Contextuality as a Heuristic for Memory Separations in Learning
• Quantum Algorithms for State Preparation and Data Classification based on Stabilizer Codes
• Detecting underdetermination in parameterized quantum circuits
• Vortex Detection from Quantum Data