量研学術機関リポジトリ「QST-Repository」は、国立研究開発法人 量子科学技術研究開発機構に所属する職員等が生み出した学術成果（学会誌発表論文、学会発表、研究開発報告書、特許等）を集積しインターネット上で広く公開するサービスです。 Welcome to QST-Repository where we accumulates and discloses the academic research results(Journal Publications, Conference presentation, Research and Development Report, Patent, etc.) of the members of National Institutes for Quantum Science and Technology.
Thank you very much for using our website. On the 11th of March 2019, this site was moved from our own network server to the JAIRO Cloud network server. If you previously bookmarked this site, that bookmark will no longer work. We would be grateful if you could bookmark the website again. Thank you very much for your understanding and cooperation.
Real experimental measurements in high-radiation environments often suffer from a high-flux of background noise which can limit the retrieval of the underlying signal. It is important to have an effective method to properly remove unwanted noise from measurement images. Machine learning methods using a multilayer neural network (deep learning) have been shown to be effective for extracting features from images. However, the efficacy of such methods is often restricted by a lack of high-quality training data. Here, we demonstrate the application for noise removal by performing simulations to generate virtual training data for a denoising deep-learning model. We first apply the model to simulations of an electron spectrometer measuring the energy spectra of electron beams accelerated from the interaction of an intense laser with a thin foil. By considering the chi-squared test and image test-indexes, namely the peak signal-to-noise ratio (PSNR) and structural similarity index measure (SSIM), we found our method to be highly effective. We then used the trained model to denoise real experimental measurements of the electron beam spectra from experiments performed at a state-of-the-art high-power laser facility. This application is offered as a new method for effectively removing noise from experimental data in high-flux radiation background environment.
雑誌名
Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment
Denoising application for electron spectrometer in laser-driven ion acceleration using a Simulation-supervised Learning based CDAE
