A5001581628- Machine Learning in Materials Science
- Geological and Geochemical Analysis
- High-pressure geophysics and materials
- Geochemistry and Geologic Mapping
- Electron and X-Ray Spectroscopy Techniques
- Graphene research and applications
- earthquake and tectonic studies
- Geomagnetism and Paleomagnetism Studies
- Polyoxometalates: Synthesis and Applications
- Hydrocarbon exploration and reservoir analysis
- Geology and Paleoclimatology Research
- Telomeres, Telomerase, and Senescence
- Advancements in Battery Materials
- Mining and Gasification Technologies
- Geophysical and Geoelectrical Methods
- Nonlinear Dynamics and Pattern Formation
- Algorithms and Data Compression
- Aerodynamics and Fluid Dynamics Research
- Advanced Semiconductor Detectors and Materials
- Electric and Hybrid Vehicle Technologies
- Geophysics and Sensor Technology
- Advanced Electron Microscopy Techniques and Applications
- Elasticity and Wave Propagation
- Mesenchymal stem cell research
- Face and Expression Recognition
National Institute of Advanced Industrial Science and Technology
2017-2024
Nankai University
2021
Xijing University
2021
Philipps University of Marburg
2021
Sichuan University of Science and Engineering
2021
Hubei University
2021
Hubei University of Arts and Science
2021
Shizuoka University
2009-2017
Institute of Geosciences
2012
Mie University
2008
Prediction of reaction yields by machine learning approach is demonstrated in tungsten-catalyzed epoxidation alkenes. The various electronic and vibrational parameters the phosphonic acids are collected DFT simulation, chosen LASSO as essential for prediction yields. With trained model, we can predict with unverified an error 26%.
We introduce a spectrum-adapted expectation-maximization (EM) algorithm for high-throughput analysis of large number spectral datasets by considering the weight intensity corresponding to measurement energy steps. Proposed method was applied synthetic data in order evaluate performance accuracy and calculation time. Moreover, proposed performed collected from graphene MoS2 field-effect transistors devices. The completed less than 13.4 s per set successfully detected systematic peak shifts C...
Monolayer transition metal dichalcogenides (TMDs) have been considered as promising materials for various next-generation semiconductor devices. However, carrier doping techniques TMDs, which are important device fabrication, not completely established yet. Here, we report a monolayer p–n junction formed using in situ substitutional during chemical vapor deposition (CVD). We synthesized MoS2–Nb-doped MoS2 lateral homojunctions CVD and then characterized their physical electrical properties....
We propose a fitting model that automatically conducts the background subtraction during high-throughput peak fitting. The consists of pseudo-Voigt mixture and ramp-sum model, each represents component, respectively. optimization is performed by spectrum adapted ECM algorithm enables us to perform simultaneously through calculation. Application proposed synthetic spectral data showed appropriate decomposition component. also applied 3721 collected from SnS sheet X-ray photoelectron...
We introduced the spectrum-adapted expectation-conditional maximization (ECM) algorithm to improve efficiency of peak fitting spectral data by various models. The ECM can perform using Pseudo–Voigt mixture model and Doniach–Šunjić–Gauss which are generally used for in X-ray photoelectron spectroscopy. Analyses synthetic experimental showed that proposed method quickly completed calculation estimated well-fitted curves data. This result suggests spectrum adapted efficiently large number sets.
Laser-induced breakdown spectroscopy (LIBS) offers a noninvasive, label-free technique for chemical analysis in challenging environments, including deep-sea mineral resource evaluation and extra-terrestrial geology. We aim to improve the usefulness of LIBS spectral these applications. propose an efficient, systematic procedure that uses calibration-free (CF-LIBS) quantitatively estimate compositions. This method combines baseline estimation denoising using sparsity with spectrum-adapted...
We introduce a peak fitting method to estimate the model parameters and number of peaks without using conventional trial-and-error approach. The proposed automatically removes excess maximum posteriori estimation. computation is performed efficiently by spectrum-adapted expectation–conditional maximisation algorithm with deterministic annealing. apply synthetic experimental data from tunnel field-effect transistor. identified two components in MoS2 sheet, which are interpreted be Mo 3d3/2...
Abstract The prediction of the initial reaction rate in tungsten-catalyzed epoxidation alkenes by using a machine learning approach is demonstrated. ensemble framework used this study consists random sampling with replacement from training dataset, construction several predictive models (weak learners), and combination their outputs. This enables us to obtain reasonable model that avoids problem overfitting, even when analyzing small dataset.
Microboudin paleopiezometry is an intensive endeavor that involves measurement of several hundred grains per sample to produce reliable estimations far–field differential stress. This procedure particularly time–consuming when conducting stress analysis for a large number samples within metamorphic belt. To improve and expedite the estimation procedure, we propose numerical model uses grain–shape data calculate relationship between proportion microboudinaged columnar (p) (σ0). Our combines...
Abstract We use Bayesian modeling of the equation state (EoS) to constrain density ( ρ ) and P wave velocity V liquid iron under conditions Earth's outer core. Experiments at such high pressures temperatures T are technically challenging, so there few data available in parameter optimization EoS. Our inference successfully estimates posterior probability distribution parameters unobserved by using Hamiltonian Monte Carlo method. These distributions allow calculation ‐ profiles along...
We carried out statistical evaluations of two probability density functions for the microboudin palaeopiezometer using Akaike information criterion (AIC) and cross-validation (CV) technique. In terms relevant stress-transfer model, these are defined as elastic matrix model Newtonian viscous respectively. The AIC CV techniques enable us to evaluate relative quality both models when applied nine data sets collected from metachert samples containing tourmaline grains in a quartz matrix, East...
Informatics techniques support improving the efficiency of data analysis in spectral imaging measurements. Applications informatics to measurement technique are categorized as informatics, which is growing parallel alongside with materials informatics. Recently, we introduced a spectrum-adapted expectation-maximization (EM) algorithm for high-throughput large number datasets obtained by synchrotron soft X-ray scanning photoelectron microscopy. The advantage proposed method that high-speed...