arXiv subjects

Yuta Sumiya

Publications and source records attributed to Yuta Sumiya.

4 recordsLinked to original sources

arXiv2026

Development of ultra-high efficiency soft X-ray angle-resolved photoemission spectroscopy equipped with deep prior-based denoising method

Soft X-ray angle resolved photoemission spectroscopy (SX-ARPES) is one of the most powerful spectroscopic techniques to visualize the three-dimensional bulk electronic structure in reciprocal lattice space. Compared with ARPES employing low-energy photon sources, the time burden imposed by a lower photoelectron yield, stemming from the photoionization cross-section, has been a persistent technical challenge. To address this challenge, we have developed a noise removal system by using the deep prior-based method and integrated it into the micro focused SX-ARPES (μSX-ARPES) system at BL25SU in SPring-8. Our implemented system effectively eliminates the grid and spike noise typically present in ARPES data acquired using the voltage Fixed-mode, within about 30 seconds. We demonstrate, through the μSX-ARPES measurements on a single crystal of CeRu2Si2, that data with sufficient statistical accuracy can be obtained in approximately 40 seconds. In addition, we present the potential of high signal-to-noise ratio ARPES measurement, achieving an energy resolution of 51.6 meV at an excitation energy of 708 eV in μSX-ARPES measurements on polycrystalline gold. Our developed system successfully reduces the time burden in SX-ARPES and paves the way for advancements in lower photoelectron yield measurements, such as those requiring higher energy resolution and three-dimensional nonequilibrium measurements.

cond-mat.str-el↗

arXiv2025

Deep prior-based denoising for state-of-the-art scientific imaging and metrology

Deep learning has revolutionized computer vision, yet a major gap persists between complex, data-hungry deep learning models and the practical demands of state-of-the-art scientific measurements. To fundamentally bridge this gap, we propose deep prior-based denoising, a robust deep learning model that requires no training data. We demonstrate its effectiveness by removing grid artifacts in angle-resolved photoemission spectroscopy (ARPES), a long-standing and critical data analysis challenge in materials science. Our results demonstrate that deep prior-based denoising yields clearer ARPES images in a fraction of the time required by conventional, experiment-based denoising methods. This ultra-efficient approach to ARPES will enable high-speed, high-resolution three-dimensional band structure mapping in momentum space, thereby dramatically accelerating our understanding of microscopic electronic structures of materials. Beyond ARPES, deep prior-based denoising represents a versatile tool that could become a new standard in any advanced scientific measurement fields where data acquisition is limited.

cond-mat.mtrl-sci↗

arXiv2024

Model-Based Counterfactual Explanations Incorporating Feature Space Attributes for Tabular Data

Machine-learning models, which are known to accurately predict patterns from large datasets, are crucial in decision making. Consequently, counterfactual explanations-methods explaining predictions by introducing input perturbations-have become prominent. These perturbations often suggest ways to alter the predictions, leading to actionable recommendations. However, the current techniques require resolving the optimization problems for each input change, rendering them computationally expensive. In addition, traditional encoding methods inadequately address the perturbations of categorical variables in tabular data. Thus, this study propose FastDCFlow, an efficient counterfactual explanation method using normalizing flows. The proposed method captures complex data distributions, learns meaningful latent spaces that retain proximity, and improves predictions. For categorical variables, we employed TargetEncoding, which respects ordinal relationships and includes perturbation costs. The proposed method outperformed existing methods in multiple metrics, striking a balance between trade offs for counterfactual explanations. The source code is available in the following repository: https://github.com/sumugit/FastDCFlow.

cs.LG↗

arXiv2023

Pseudo Session-Based Recommendation with Hierarchical Embedding and Session Attributes

Recently, electronic commerce (EC) websites have been unable to provide an identification number (user ID) for each transaction data entry because of privacy issues. Because most recommendation methods assume that all data are assigned a user ID, they cannot be applied to the data without user IDs. Recently, session-based recommendation (SBR) based on session information, which is short-term behavioral information of users, has been studied. A general SBR uses only information about the item of interest to make a recommendation (e.g., item ID for an EC site). Particularly in the case of EC sites, the data recorded include the name of the item being purchased, the price of the item, the category hierarchy, and the gender and region of the user. In this study, we define a pseudo--session for the purchase history data of an EC site without user IDs and session IDs. Finally, we propose an SBR with a co-guided heterogeneous hypergraph and globalgraph network plus, called CoHHGN+. The results show that our CoHHGN+ can recommend items with higher performance than other methods.

cs.IR↗