A5063091303- Machine Learning in Materials Science
- Biomedical Text Mining and Ontologies
- High Temperature Alloys and Creep
- Topic Modeling
- Semantic Web and Ontologies
- Advanced Materials Characterization Techniques
- Web Data Mining and Analysis
- Advanced Text Analysis Techniques
- Scientific Computing and Data Management
- Computational Drug Discovery Methods
- Microstructure and Mechanical Properties of Steels
- Aluminum Alloy Microstructure Properties
- Manufacturing Process and Optimization
- Graphene research and applications
- scientometrics and bibliometrics research
- Software Engineering Research
- Research Data Management Practices
- Magnetic and transport properties of perovskites and related materials
- Data Quality and Management
- Academic Writing and Publishing
- Multiferroics and related materials
- Advanced X-ray Imaging Techniques
- Advanced Computing and Algorithms
- Carbon dioxide utilization in catalysis
- Chemical and Physical Properties of Materials
National Institute for Materials Science
2017-2025
Soochow University
2020
The University of Tokyo
2017-2019
RIKEN Center for Advanced Intelligence Project
2018
Hokkaido University
2011-2016
The automatic extraction of chemical information from text requires the recognition entity mentions as one its key steps. When developing supervised named (NER) systems, availability a large, manually annotated corpus is desirable. Furthermore, large corpora permit robust evaluation and comparison different approaches that detect chemicals in documents. We present CHEMDNER corpus, collection 10,000 PubMed abstracts contain total 84,355 labeled by expert chemistry literature curators,...
We investigated oxygen reduction reaction (ORR) properties of Pt-containing compositionally complex alloy (Pt-CCA) single-crystal model catalyst surfaces to optimize dry-process synthesis conditions, that is, CCA compositions less-noble alloying elements and their (annealing) temperatures. Using a machine-learning approach, we effectively navigated the large space possible conditions minimize experimental workload. The ORR activity durability Pt/CCA/Pt(111) (synthesized through vacuum...
Heteroatom doping has endowed graphene with manifold aspects of material properties and boosted its applications. The atomic structure determination doped is vital to understand properties. Motivated by the recently synthesized boron-doped relatively high concentration, here we employ machine learning methods search most stable structures boron atoms in graphene, conjunction atomistic simulations. From determined structures, find that free-standing pristine energetically prefer substitute...
Complex materials design is often represented as a black-box combinatorial optimization problem. In this paper, we present novel python library called MDTS (Materials Design using Tree Search). Our algorithm employs Monte Carlo tree search approach, which has shown exceptional performance in computer Go game. Unlike evolutionary algorithms that require user intervention to set parameters appropriately, no tuning and works autonomously various problems. comparison Bayesian package, our showed...
Abstract Materials design and discovery can be represented as selecting the optimal structure from a space of candidates that optimizes target property. Since number exponentially proportional to determination variables, must obtained efficiently. Recently, inspired by its success in Go computer game, several approaches have applied Monte Carlo tree search (MCTS) solve optimization problems natural sciences including materials science. In this paper, we briefly reviewed applications MCTS...
The G7 Summit in 2023 Promoted Open Science, Leading to the Practical Implementation of FAIRable Research Data Infrastructures and Applications. This Has Increased Researchers' Awareness Interoperable Digital Transformation. Focus on Battery Materials Aims Support Sustainable Development Goals (SDGs), an Important Global Issue. Using Materials-Related Papers, Key Terms Were Extracted with KeyBERT, Frequent Term Analysis Was Conducted. Author Similarity Connected Researchers Similar...
A growing number of papers are published in the area superconducting materials science. However, novel text and data mining (TDM) processes still needed to efficiently access exploit this accumulated knowledge, paving way towards data-driven design. Herein, we present SuperMat (Superconductor Materials), an annotated corpus linked derived from scientific publications on superconductors, which comprises 142 articles, 16052 entities, 1398 links that characterised into six categories: names,...
Data-driven machine learning (ML) has earned remarkable achievements in accelerating materials design, while it heavily relies on high-quality data acquisition. In this work, we develop an adaptive design framework for searching optimal starting from zero and with as few DFT calculations possible. This integrates automatic density functional theory (DFT) improved Monte Carlo tree search via reinforcement algorithm (MCTS-PG). As a successful example, apply to rapidly identify the desired...
In this paper, a state-of-the-art Artificial Intelligence (AI) technique is used for precipitation hardening of Ni-based alloy to predict more flexible non-isothermal aging (NIA) and examine the possible routes enhancement in strength that may be practically achieved. Additionally, AI integrate with Materials Integration by Network Technology, which computational workflow utilized model microstructure evolution evaluate 0.2% proof stress isothermal NIA. As result, it find enhanced NIA fixed...
To support nanocrystal device development, we have been working on a computational framework to utilize information in research papers devices. We developed an annotated corpus called “ NaDev” ( Na nocrystal Dev ice Development) for this purpose. also proposed automatic extraction system “NaDevEx” Automatic Information Ex traction Framework). NaDevEx aims at extracting from devices using the NaDev and machine-learning techniques. However, characteristics of were not examined detail. In...
We present a general machine-learning-based approach to solve the inverse design problem of depth-graded multilayer structures (so-called supermirrors) for x-ray optics. Our model uses Monte Carlo tree search (MCTS) with policy gradient in combination reflectivity simulation. MCTS is an iterative method that showed competitive efficiency materials and discovery problems. A algorithm neural network was added optimize expansion. The reinforcement learning optimizes parametrized policies toward...
To support development processes for nanodevices, we want to utilize the information related experiments from nanodevice research papers. In this paper, propose a new guideline constructing tagged corpus of papers achieve goal. We also use construction tool. speed up process, an approach automatic extraction experiment-related
Nanodevices development process is not well systematized. In order to support this process, we have been working on an experimental record management system that aims at analyzing previous manufacturing experiments provide insight future manufacturing. However, found records do enough information for detailed analysis. paper, propose approach extract extra (metadata) from papers can enhance the analysis of results.
In utilizing nanodevice development research papers to assist in experimental planning and design, it is useful identify annotate characteristic categories of information contained those such as source material, evaluation parameter, etc.In order support this annotation process, we have been working construct a corpus complementary automatic scheme.Due the variations terms, however, recall some was not adequate.In paper, propose use basic physical quantities list extract parameter...
The process of nanocrystal device development is not well systematized. To support this process, analysis the information produced by developmental experiments required. In study, we constructed an annotated corpus to extraction experimental from relevant publications. We designed corpus-construction guidelines cooperating with a domain expert. evaluated these through graduate students domain, and then construction experiments, achieved sufficient level Inter-Annotator Agreement using loose...
Battery related research has gained increasing attention in recent years. A large number of papers about battery are being published as open access. In order to explore the content these efficiently, it is necessary develop information management tools capture knowledge embedded them. We working on a project create visual topic map for researchers. This connects researchers according their topics. allows potential connection between researches with similar topics, which increase...
<title>Abstract</title> This study presents the comprehensive analysis of flexible non-isothermal aging (NIA) patterns discovered through artificial intelligence (AI) to improve mechanical strength (0.2% proof stress) in γ – γ' two-phase, binary Ni-Al alloys. In our recent investigation, we found that AI algorithm could propose with superior compared conventional isothermal aging. this current study, continued extensive exploration methodologies, uncovering diverse also surpassed benchmark....
As an alternative to conventional plastic dishes, the interface between water-immiscible hydrophobic fluids, such as perfluorocarbons and silicones, permits cell adhesion growth. Thus, it is expected replace petroleum-derived products in a sustainable society. However, most fluids are cytotoxic, which limits range of mechanical chemical cues exposed cells. Using data-driven approach, this study aimed identify non-cytotoxic ionic liquids (ILs) fluid culture platforms take advantage their...
SAMURAI (NIMS 2022), a directory service of the National Institute for Materials Science (NIMS) researchers in Japan was launched 2009 following development NIMS institutional repository (Tanifuji et al. 2019). The concept is to synchronize between profile information and their publications which are self-archived system. renewed 2017 with interoperable functions ORCID. supports various links not only individual articles patents, but also databases such as KAKEN (Database Grants-in-Aid...
Depth-graded multilayer structures are widely used in X-ray related applications. In this paper, we propose an optimization approach using machine learning principles to accelerate depth-graded design. We use Monte Carlo tree search (MCTS) find optimal thickness for each layer the structure that achieves maximum mean reflectivity angular range at a specific beam energy. obtained 0.78 0.4~0.55° Cu Kα radiation approach. For top structure, could achieve small standard deviation of 0.016 within...