Quantum Neural Fields

ICLR 2025 Conference Submission348 Authors

Keywords: quantum neural fields, representation learning, Neuro-deterministic data encoding, quantum ansatz

TL;DR: quantum neural fields for representing 2D images and 3D shapes that are compatible with gate-based quantum computers

Abstract: This paper introduces a new type of neural field for visual computing with components compatible with gate-based quantum hardware or simulators thereof. Our Quantum Neural Field Network (QNF-Net) expects as input a query coordinate and, optionally, a latent variable value, and outputs the corresponding field value. QNF-Net includes a new feature map for classical data encoding and a parametrised quantum circuit. The proposed neuro-deterministic data encoding converts, into qubit amplitudes, an energy spectrum of the Gibbs-Boltzmann distribution corresponding to the learned problem energy manifold. We provide a theoretical analysis of the model and its components and perform experiments on a simulator of a gate-based quantum computer with 2D images and 3D shapes (and their collections as learnt priors) and compare results with several classical baselines. QNF-Net consistently outperforms the classical baselines with a comparable number of parameters and achieves faster convergence speed, therefore showing its potential quantum advantages, even for relatively large-scale problems compared to what has been demonstrated in quantum machine learning so far. We will release the source code to facilitate method reproducibility.

Primary Area: unsupervised, self-supervised, semi-supervised, and supervised representation learning

Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.

Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2025/AuthorGuide.

Reciprocal Reviewing: I understand the reciprocal reviewing requirement as described on https://iclr.cc/Conferences/2025/CallForPapers. If none of the authors are registered as a reviewer, it may result in a desk rejection at the discretion of the program chairs. To request an exception, please complete this form at https://forms.gle/Huojr6VjkFxiQsUp6.

Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors’ identity.

No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.

Submission Number: 348