检索结果-南通市图书馆

A critical examination of robustness and generalizability of machine learning prediction of materials properties

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

npj Computational Materials 2023年第1期9卷 1787-1795页

作者： Kangming Li Brian DeCost kamal choudhary Michael Greenwood Jason Hattrick-Simpers Department of Materials Science and Engineering University of Toronto27 King’s College CirTorontoONCanada Material Measurement Laboratory National Institute of Standards and Technology100 Bureau DrGaithersburgMDUSA Theiss Research La JollaCA 92037USA Canmet MATERIALS Natural Resources Canada183 Longwood Road southHamiltonONCanada

Recent advances in machine learning(ML)have led to substantial performance improvement in material database benchmarks,but an excellent benchmark score may not imply good generalization *** we show that ML models trained on Materials Project 2018 can have severely degraded performance on new compounds in Materials Project 2021 due to the distribution *** discuss how to foresee the issue with a few simple ***,the uniform manifold approximation and projection(UMAP)can be used to investigate the relation between the training and test data within the feature ***,the disagreement between multiple ML models on the test data can illuminate out-of-distribution *** demonstrate that the UMAP-guided and query by committee acquisition strategies can greatly improve prediction accuracy by adding only 1%of the test *** believe this work provides valuable insights for building databases and models that enable better robustness and generalizability.

关键词： prediction adding critical

Efficiently searching extreme mechanical properties via boundless objective-free exploration and minimal first-principles calculations

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

npj Computational Materials 2022年第1期8卷 1336-1347页

作者： Joshua Ojih Mohammed Al-Fahdi Alejandro David Rodriguez kamal choudhary Ming Hu Department of Mechanical Engineering University of South CarolinaColumbiaSC29208USA Materials Science and Engineering Division National Institute of Standards and TechnologyGaithersburgMD20899USA

Despite the machine learning(ML)methods have been largely used recently,the predicted materials properties usually cannot exceed the range of original training *** deployed a boundless objective-free exploration approach to combine traditional ML and density functional theory(DFT)in searching extreme material *** combination not only improves the efficiency for screening large-scale materials with minimal DFT inquiry,but also yields properties beyond original training *** use Stein novelty to recommend outliers and then verify using *** data are then added into the training dataset for next round *** test the loop of training-recommendation-validation in mechanical property *** screening 85,707 crystal structures,we identify 21 ultrahigh hardness structures and 11 negative Poisson’s ratio *** algorithm is very promising for future materials discovery that can push materials properties to the limit with minimal DFT calculations on only~1%of the structures in the screening pool.

关键词： extreme properties minimal

An AI-driven microstructure optimization framework for elastic properties of titanium beyond cubic crystal systems

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

npj Computational Materials 2023年第1期9卷 1217-1226页

作者： Yuwei Mao Mahmudul Hasan Arindam Paul Vishu Gupta kamal choudhary Francesca Tavazza Wei-keng Liao Alok choudhary Pinar Acar Ankit Agrawal Department of Electrical and Computer Engineering Northwestern UniversityEvanstonILUSA Virginia Tech BlacksburgVAUSA Materials Measurement Laboratory National Institute of Standards and TechnologyGaithersburgMD 20899USA Theiss Research La JollaCA 92037USA

Materials design aims to identify the material features that provide optimal properties for various engineering applications,such as aerospace,automotive,and *** of the important but challenging problems for materials design is to discover multiple polycrystalline microstructures with optimal *** paper proposes an end-to-end artificial intelligence(AI)-driven microstructure optimization framework for elastic properties of *** this work,the microstructure is represented by the Orientation Distribution Function(ODF)that determines the volume densities of crystallographic *** framework was evaluated on two crystal systems,cubic and hexagonal,for Titanium(Ti)in Joint Automated Repository for Various Integrated Simulations(JARVIS)database and is expected to be widely applicable for materials with multiple crystal *** proposed framework can discover multiple polycrystalline microstructures without compromising the optimal property values and saving significant computational time.

关键词： microstructure cubic crystal

Author Correction:Atomistic Line Graph Neural Network for improved materials property predictions

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

npj Computational Materials 2022年第1期8卷 2117-2118页

作者： kamal choudhary Brian DeCost Materials Measurement Laboratory National Institute of Standards and Technology Gaithersburg MD 20899 USA Theiss ResearchLa Jolla California CA 92037 USA DeepMaterials LLC Silver Spring MD 20906 USA

The original version of this Article contained errors in values of ALIGNN data in Table *** a result,the following changes have been made to the original version of this Article:In Table 5,the data for“OrbNetens5”co... 详细信息

关键词： property removed column

Atomistic Line Graph Neural Network for improved materials property predictions

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

npj Computational Materials 2021年第1期7卷 1691-1698页

作者： kamal choudhary Brian DeCost Materials Measurement Laboratory National Institute of Standards and TechnologyGaithersburgMD 20899USA Theiss ResearchLa Jolla California 92037USA DeepMaterials LLC Silver SpringMD 20906USA

Graph neural networks(GNN)have been shown to provide substantial performance improvements for atomistic material representation and modeling compared with descriptor-based machine learning *** most existing GNN models for atomistic predictions are based on atomic distance information,they do not explicitly incorporate bond angles,which are critical for distinguishing many atomic ***,many material properties are known to be sensitive to slight changes in bond *** present an Atomistic Line Graph Neural Network(ALIGNN),a GNN architecture that performs message passing on both the interatomic bond graph and its line graph corresponding to bond *** demonstrate that angle information can be explicitly and efficiently included,leading to improved performance on multiple atomistic prediction *** ALIGNN models for predicting 52 solid-state and molecular properties available in the JARVIS-DFT,Materials project,and QM9 *** can outperform some previously reported GNN models on atomistic prediction tasks by up to 85%in accuracy with better or comparable model training speed.

关键词： explicitly property passing

Million-scale data integrated deep neural network for phonon properties of heuslers spanning the periodic table

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

npj Computational Materials 2023年第1期9卷 2155-2166页

作者： Alejandro Rodriguez Changpeng Lin Hongao Yang Mohammed Al-Fahdi Chen Shen kamal choudhary Yong Zhao Jianjun Hu Bingyang Cao Hongbin Zhang Ming Hu Department of Mechanical Engineering University of South CarolinaColumbiaSC 29208USA Theory and Simulation of Materials(THEOS) École Polytechnique Fédérale de LausanneCH-1015 LausanneSwitzerland Key Laboratory for Thermal Science and Power Engineering of Ministry of Education Department of Engineering MechanicsTsinghua UniversityBeijing 100084China Institute of Materials Science Technical University of DarmstadtDarmstadt 64287Germany Materials Science and Engineering Division National Institute of Standards and TechnologyGaithersburgMD 20899USA Department of Computer Science and Engineering University of South CarolinaColumbiaSC 29208USA

Existing machine learning potentials for predicting phonon properties of crystals are typically limited on a material-to-materialbasis, primarily due to the exponential scaling of model complexity with the number of atomic species. We address this bottleneckwith the developed Elemental Spatial Density Neural Network Force Field, namely Elemental-SDNNFF. The effectiveness andprecision of our Elemental-SDNNFF approach are demonstrated on 11,866 full, half, and quaternary Heusler structures spanning 55elements in the periodic table by prediction of complete phonon properties. Self-improvement schemes including active learningand data augmentation techniques provide an abundant 9.4 million atomic data for training. Deep insight into predicted ultralowlattice thermal conductivity (<1 Wm^(−1) K^(−1)) of 774 Heusler structures is gained by p–d orbital hybridization analysis. Additionally, aclass of two-band charge-2 Weyl points, referred to as “double Weyl points”, are found in 68% and 87% of 1662 half and 1550quaternary Heuslers, respectively.

关键词： properties phonon spanning

Computational search for magnetic and non-magnetic 2D topological materials using unified spin-orbit spillage screening

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

npj Computational Materials 2020年第1期6卷 1272-1279页

作者： kamal choudhary Kevin F.Garrity Jie Jiang Ruth Pachter Francesca Tavazza Materials Science and Engineering Division National Institute of Standards and TechnologyGaithersburgMD 20899USA Materials Directorate Air Force Research LaboratoryWright–Patterson Air Force BaseOH 45433USA

Two-dimensional topological materials(2D TMs)have a variety of properties that make them attractive for applications including spintronics and quantum ***,there are only a few such experimentally known *** help discover new 2D TMs,we develop a unified and computationally inexpensive approach to identify magnetic and non-magnetic 2D TMs,including gapped and semi-metallic topological classifications,in a high-throughput way using density functional theory-based spin–orbit spillage,Wannier-interpolation,and related techniques.

关键词： materials unified topological

Temporal and spatial compliance behaviour of pedestrians under the influence of time pressure at signalized intersections:A pedestrian simulator study

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Journal of Traffic and Transportation Engineering(English Edition) 2024年第1期11卷 55-68页

作者： Apurwa Dhoke Pushpa choudhary Department of Civil Engineering Indian Institute of Technology RoorkeeRoorkee 247667India

Pedestrian safety is at high stakes due to the non-compliance practices of pedestrians at signalized ***,when pedestrians are hurrying,they deliberately engage in such unsafe ***,the purpose of this study was to understand how time pressure(i.e.,feeling of hurry or saving time)affected pedestrians'decisions to follow traffic rules at signalized *** achieve the study objectives,a pedestrian simulator setup was used to collect the crossing behaviour of forty participants at a four-legged signalized ***-compliance,one of the riskiest pedestrian behaviours,was examined with respect to three different forms,comprising dangerous temporal non-compliance(D-TNC),non-dangerous temporal non-compliance(ND-TNC),and spatial non-compliance(SNC)behaviour under two distinct conditions:baseline(i.e.,no time pressure)and time pressure *** effects of demographics,usual walking features,and time pressure on D-TNC and ND-TNC were investigated using a multinomial regression model,while SNC behaviour was investigated using a binary regression *** was interesting to note that the majority of the factors related to pedestrians’usual walking behaviour had an impact on all kinds of non-compliance ***,the results also showcased that time pressure had a contrasting impact on D-TNC and ND-TNC behaviour whereas SNC behaviour increased under time ***,the varying impacts of D-TNC,ND-TNC,and SNC were also reflected in the occurrence of the crashes,which were probably triggered by discrepancies in the influence of time pressure on non-compliance *** findings highlight the need for technical solutions,educational outreach,and efficient enforcement practices to reduce pedestrians'non-compliant behaviour.

关键词： Signalized intersection Time pressure Pedestrians Temporal compliance Spatial compliance Pedestrian simulator

Recent advances and applications of deep learning methods in materials science

维普期刊数据库

同方期刊数据库评论

在线全文

学校读者我要写书评

暂无评论

npj Computational Materials 2022年第1期8卷 548-573页

作者： kamal choudhary Brian DeCost Chi Chen Anubhav Jain Francesca Tavazza Ryan Cohn Cheol Woo Park Alok choudhary Ankit Agrawal Simon J.L.Billinge Elizabeth Holm Shyue Ping Ong Chris Wolverton Materials Science and Engineering Division National Institute of Standards and TechnologyGaithersburgMD20899USA Theiss Research La JollaCA92037USA DeepMaterials LLC Silver SpringMD20906USA Material Measurement Science Division National Institute of Standards and TechnologyGaithersburgMD20899USA Department of NanoEngineering University of California San DiegoSan DiegoCA92093USA Energy Technologies Area Lawrence Berkeley National LaboratoryBerkeleyCAUSA Department of Materials Science and Engineering Carnegie Mellon UniversityPittsburghPA15213USA Department of Materials Science and Engineering Northwestern UniversityEvanstonIL60208USA Department of Electrical and Computer Engineering Northwestern UniversityEvanstonIL60208USA Department of Applied Physics and Applied Mathematics and the Data Science Institute Fu Foundation School of Engineering and Applied SciencesColumbia UniversityNew YorkNY10027USA

Deep learning(DL)is one of the fastest-growing topics in materials data science,with rapidly emerging applications spanning atomistic,image-based,spectral,and textual data *** allows analysis of unstructured data and automated identification of *** recent development of large materials databases has fueled the application of DL methods in atomistic prediction in *** contrast,advances in image and spectral data have largely leveraged synthetic data enabled by high-quality forward models as well as by generative unsupervised DL *** this article,we present a high-level overview of deep learning methods followed by a detailed discussion of recent developments of deep learning in atomistic simulation,materials imaging,spectral analysis,and natural language *** each modality we discuss applications involving both theoretical and experimental data,typical modeling approaches with their strengths and limitations,and relevant publicly available software and *** conclude the review with a discussion of recent cross-cutting work related to uncertainty quantification in this field and a brief perspective on limitations,challenges,and potential growth areas for DL methods in materials science.

关键词： learning limitations textual