Solving the Coloring Problem in Half-Heusler Structures: Machine-Learning Predictions and Experimental Validation

Gzyl, Alexander; Oliynyk, Anton O.; Adutwum, Lawrence A.; Mar, Arthur

doi:10.1021/acs.inorgchem.9b00987

Cited by 21 publications

(14 citation statements)

References 51 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…An optimized trade-off between exploration (high uncertainty regions) and exploitation (best predicted performance regions), was achieved by experimentally comparing multiple design strategies. Thus, they were able to produce an optimal criterion for the synthesis of the piezoelectric (Ba [7], ANN [30], RF [12], decision tree (DT) [21,30], recursive feature elimination (RFE) [37], least absolute shrinkage and selection operator (LASSO) [2], cluster resolution feature selection (CR-FS) [7,8,19], DoE [3], LR [12,21,30], PR [30], partial least squares (PLS) [7], matrix-based recommender [48], synthetic minority oversampling technique (SMOTE) [7],…”

Section: Non-bayesian Optimization (Bo)mentioning

confidence: 99%

“…Once SMOTE had been applied, use of an ensemble approach increased the validation set sensitivity (rate of true positives) from 83.3 % (best individual model, SVM CR-FS) to 88.3 % while maintaining near identical validation specificity and accuracy. Gzyl et al [8] used 179 experimentally reported structures, 23 descriptors (selected via CR-FS from 243 descriptors based on 43 elemental properties), and SVM to classify half-Heusler site preferences resulting in a sensitivity, selectivity, and accuracy of 93%, 96%, and 95%, respectively. One goal of the work was to apply data sanitation by retesting classified candidates with various classification probabilities.…”

Section: Support Vector Machine (Svm) and Cluster Resolution Feature ...mentioning

confidence: 99%

See 1 more Smart Citation

Data-Driven Materials Discovery and Synthesis using Machine Learning Methods

Baird¹,

Marianne²,

Sayeed³

et al. 2022

Preprint

View full text Add to dashboard Cite

Experimentally and computationally [39][40][41][42][43][44][45][46][47][48][49][50] validated machine learning (ML) articles are sorted based on size of the training data: 1-100, 101-10 000, and 10 000+ in a comprehensive set summarizing legacy and recent advances in the field. The review emphasizes the interrelated fields of synthesis, characterization, and prediction. Size range 1-100 consists mostly of Bayesian optimization (BO) articles, whereas 101-10 000 consists mostly of support vector machine (SVM) articles. The articles often use combinations of ML, feature selection (FS), adaptive design (AD), high-throughput (HiTp) techniques, and domain knowledge to enhance predictive performance and/or model interpretability. Grouping cross-validation (G-CV) techniques curb overly optimistic extrapolative predictive performance. Smallerdatasets relying on AD are typically able to identify new materials with desired properties but do so in a constrained design space. In larger datasets, the low-hanging fruit of materials optimization are typically already discovered, and the models are generally less successful at extrapolating to new mate-

show abstract

Section: Non-bayesian Optimization (Bo)mentioning

confidence: 99%

Section: Support Vector Machine (Svm) and Cluster Resolution Feature ...mentioning

confidence: 99%

Data-Driven Materials Discovery and Synthesis using Machine Learning Methods

Baird¹,

Marianne²,

Sayeed³

et al. 2022

Preprint

View full text Add to dashboard Cite

show abstract

“…Anton Oliynyk's research: The Oliynyk group studies intermetallic compounds and combines machine learning methods with experimental research. [ 490–495 ] They study the inorganic chemistry of intermetallic materials with focus on energy‐converting materials and mechanical properties such as hardness and wear resistance. [ 493 ] The materials are synthesized directly from elements by high‐temperature methods, including arc‐melting and sintering.…”

Section: Overview Of Mercury Faculty Research Effortsmentioning

confidence: 99%

Twenty years of exceptional success: The molecular education and research consortium in undergraduate computational chemistry (MERCURY)

Shields

2020

Int J of Quantum Chemistry

View full text Add to dashboard Cite

The molecular education and research consortium in undergraduate computational chemistry (MERCURY) consortium, established in 2000, has contributed greatly to the scientific development of faculty and undergraduates. The MERCURY faculty peer-reviewed publication rate from 2001 to 2019 of 1.7 papers/faculty/year is 3.4 times the rate of the physical science faculty at primarily undergraduate institutions. We have worked with over 1000 students on research projects since 2001, and 75% of our undergraduate research students have been under-represented in chemistry, either female or students of color. Approximately half of our alumni attend graduate school for the purpose of obtaining advanced degrees in STEM fields, and two-thirds are female and/or students of color. We have had more than 1600 attendees at 18 MERCURY conferences, including 111 invited speakers, 61 of whom have been female and/or faculty of color. In this paper, the research accomplishments, transformational outcomes, and scientific productivity of the MERCURY faculty are highlighted.

show abstract

“…Machine learning models have proven to be useful in identifying crystallographic information of ternary equiatomic rare earth compounds [27,28]. These ML efforts, however, are often limited in their ability to describe true thermodynamic behavior of new rare earth compounds or compositions.…”

Section: Introductionmentioning

confidence: 99%

Machine-learning enabled thermodynamic model for the design of new rare-earth compounds

Singh,

Del Rose,

Vazquez

et al. 2022

Preprint

View full text Add to dashboard Cite

We employ a descriptor based machine-learning approach to assess the effect of chemical alloying on formationenthalpy of rare-earth intermetallics. Application of machine-learning approaches in rare-earth intermetallic design have been sparse due to limited availability of reliable datasets. In this work, we developed an 'in-house' rare-earth database with more than 600+ compounds, each entry was populated with formation enthalpy and related atomic features using high-throughput density-functional theory (DFT). The SISSO (sure independence screening and sparsifying operator) based machine-learning method with meaningful atomic features was used for training and testing the formation enthalpies of rare earth compounds. The complex lattice function coupled with the machine-learning model was used to explore the effect of transition metal alloying on the energy stability of Ce based cubic Laves phases (MgCu 2 type). The SISSO predictions show good agreement with high-fidelity DFT calculations and X-ray powder diffraction measurements. Our study provides quantitative guidance for compositional considerations within a machine-learning model and discovering new metastable materials. The electronic-structure of Ce-Fe-Cu based compound was also analyzed in-depth to understand the electronic origin of phase stability. The interpretable analytical models in combination with density-functional theory and experiments provide a fast and reliable design guide for discovering technologically useful materials.

show abstract

Solving the Coloring Problem in Half-Heusler Structures: Machine-Learning Predictions and Experimental Validation

Cited by 21 publications

References 51 publications

Data-Driven Materials Discovery and Synthesis using Machine Learning Methods

Data-Driven Materials Discovery and Synthesis using Machine Learning Methods

Twenty years of exceptional success: The molecular education and research consortium in undergraduate computational chemistry (MERCURY)

Machine-learning enabled thermodynamic model for the design of new rare-earth compounds

Contact Info

Product

Resources

About