Weak-signal extraction enabled by deep-neural-network denoising of diffraction data
Abstract
Removal or cancellation of noise has wide-spread applications for imaging and acoustics. In every-day-life applications, denoising may even include generative aspects, which are unfaithful to the ground truth. For scientific use, however, denoising must reproduce the ground truth accurately. Here, we show how data can be denoised via a deep convolutional neural network such that weak signals appear with quantitative accuracy. In particular, we study X-ray diffraction on crystalline materials. We demonstrate that weak signals stemming from charge ordering, insignificant in the noisy data, become visible and accurate in the denoised data. This success is enabled by supervised training of a deep neural network with pairs of measured low- and high-noise data. We demonstrate that using artificial noise does not yield such quantitatively accurate results. Our approach thus illustrates a practical strategy for noise filtering that can be applied to challenging acquisition problems.
Cite
@article{arxiv.2209.09247,
title = {Weak-signal extraction enabled by deep-neural-network denoising of diffraction data},
author = {Jens Oppliger and M. Michael Denner and Julia Küspert and Ruggero Frison and Qisi Wang and Alexander Morawietz and Oleh Ivashko and Ann-Christin Dippel and Martin von Zimmermann and Izabela Biało and Leonardo Martinelli and Benoît Fauqué and Jaewon Choi and Mirian Garcia-Fernandez and Ke-Jin Zhou and Niels B. Christensen and Tohru Kurosawa and Naoki Momono and Migaku Oda and Fabian D. Natterer and Mark H. Fischer and Titus Neupert and Johan Chang},
journal= {arXiv preprint arXiv:2209.09247},
year = {2024}
}
Comments
14 pages, 10 figures; extended study, additional supplementary information, results unchanged