2021
DOI: 10.1038/s41524-021-00669-4
|View full text |Cite
|
Sign up to set email alerts
|

Chemical hardness-driven interpretable machine learning approach for rapid search of photocatalysts

Abstract: Strategies combining high-throughput (HT) and machine learning (ML) to accelerate the discovery of promising new materials have garnered immense attention in recent years. The knowledge of new guiding principles is usually scarce in such studies, essentially due to the ‘black-box’ nature of the ML models. Therefore, we devised an intuitive method of interpreting such opaque ML models through SHapley Additive exPlanations (SHAP) values and coupling them with the HT approach for finding efficient 2D water-splitt… Show more

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
2
1

Citation Types

1
54
0

Year Published

2022
2022
2024
2024

Publication Types

Select...
8
1

Relationship

0
9

Authors

Journals

citations
Cited by 75 publications
(55 citation statements)
references
References 85 publications
1
54
0
Order By: Relevance
“…If the outputs are discrete targets, the classification method is used to find the prediction function, whereas the regression method is used when the objective attribute is a continuous quantity (e.g., glass transition temperature). Typical classification algorithms are random forest (RF) and decision tree (DT), and regression algorithms are artificial neural networks (ANNs), support vector machine (SVM), and Gaussian process regression (GPR). Unsupervised learning seeks to detect the relationship among the input data itself, whereas supervised learning aims to create a function mapping a collection of input data to the same output attribute. Clustering is a general unsupervised learning strategy that involves splitting a data set into multiple categories, with data points in the same group or cluster being more comparable to those in other clusters.…”
Section: Machine Learning In Polymer Informaticsmentioning
confidence: 99%
“…If the outputs are discrete targets, the classification method is used to find the prediction function, whereas the regression method is used when the objective attribute is a continuous quantity (e.g., glass transition temperature). Typical classification algorithms are random forest (RF) and decision tree (DT), and regression algorithms are artificial neural networks (ANNs), support vector machine (SVM), and Gaussian process regression (GPR). Unsupervised learning seeks to detect the relationship among the input data itself, whereas supervised learning aims to create a function mapping a collection of input data to the same output attribute. Clustering is a general unsupervised learning strategy that involves splitting a data set into multiple categories, with data points in the same group or cluster being more comparable to those in other clusters.…”
Section: Machine Learning In Polymer Informaticsmentioning
confidence: 99%
“…In recent years, advances in DFT calculations have proven to be very helpful in guiding the selection or design of catalysts for efficient hydrogen generation, significantly accelerating the development of new catalysts [31][32][33][34] . For instance, Zhou et al 35 investigated various adsorption geometries of H*, HO*, O*, and H*/HO* on CdS and ZnS (110) surfaces, and explained the experimental fact that bare CdS is not an active photocatalyst for water splitting, but that Pt co-catalyst could significantly improve the catalytic activity and HER efficiency.…”
Section: Introductionmentioning
confidence: 99%
“…In recent years, advances in DFT calculations have proven to be very helpful in guiding the selection or design of photocatalysts for efficient hydrogen generation, significantly accelerating the development of new catalyst [31][32][33][34] . For instance, Zhou et al 35 investigated various adsorption geometries of H*, HO*, O* and H*/HO* on CdS and ZnS (110) surfaces, and explained the experimental fact that bare CdS is not an active photocatalyst for water splitting, but that Pt cocatalyst could significantly improve the catalytic activity and HER efficiency.…”
Section: Introductionmentioning
confidence: 99%