Predicting glass properties by using physics- and chemistry-informed machine learning models

Shih, Yueh-Ting; Shi, Yunfeng; Huang, Liping

doi:10.1016/j.jnoncrysol.2022.121511

Cited by 12 publications

(5 citation statements)

References 35 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The use of feature importance to improve the interpretability of ML models has already been used by Shih et al. in a study of physics‐ and chemistry‐informed ML models 107 …”

Section: Discussionmentioning

confidence: 99%

Enhancing glass property predictions through ab initio‐derived descriptors

Arendt,

Limbach,

Wondraczek

et al. 2024

J Am Ceram Soc.

View full text Add to dashboard Cite

The performance of ab initio descriptors derived from density functional theory simulations is systematically investigated in comparison to traditional compositional descriptors for the ability to predict glass properties utilizing machine learning algorithms. Two datasets are used for this purpose: an extensive, publicly available database involving a wide range of oxide glasses, and a small in‐house dataset covering a broader collection of inorganic glasses from metallic to non‐metallic materials. For the larger dataset, it was demonstrated that ab initio descriptors offer a substantial reduction in input dimensionality while retaining nearly equivalent predictive performance when compared to the compositional descriptors. The combination of ab initio and compositional descriptors showed an improvement in prediction accuracy. For the smaller dataset, the ab initio‐derived descriptors performed significantly better than the compositional descriptors, providing a valuable tool to improve glass property prediction in settings where the availability of data is limited. Furthermore, ab initio‐derived descriptors are not only computationally inexpensive and allow extrapolation beyond the training composition space but also facilitate model interpretation.

show abstract

“…The use of feature importance to improve the interpretability of ML models has already been used by Shih et al. in a study of physics‐ and chemistry‐informed ML models 107 …”

Section: Discussionmentioning

confidence: 99%

Enhancing glass property predictions through ab initio‐derived descriptors

Arendt,

Limbach,

Wondraczek

et al. 2024

J Am Ceram Soc.

View full text Add to dashboard Cite

show abstract

“…In the context of chemistry and materials science, ML can be employed to elucidate structure-property relationships and to determine the physical and chemical effects that govern materials properties. [83][84][85] We briefly review key ML terminology and concepts here. For an excellent detailed review of ML in the context of soft materials, the reader is directed to a topical review by Ferguson.…”

Section: Fundamentals Of Machine Learningmentioning

confidence: 99%

Beyond nature's base pairs: machine learning-enabled design of DNA-stabilized silver nanoclusters

Mastracco

Copp

2023

Chem. Commun.

View full text Add to dashboard Cite

show abstract

“…In addition to density, ML techniques have also been increasingly used to predict other glass properties, including elastic properties, glass transition temperature, coefficient of thermal expansion, liquidus temperature, refractive index, and chemical durability. [25][26][27][28] Synthesizing across the aforementioned studies 5,7,10,14,15,17,18,[20][21][22][23][24]27 reveals that there is still a lack of ML studies focusing on density prediction for silicate-based glasses that are relevant to cement and concrete applications. For the production of concrete, industrial waste rich in glassy amorphous phases (e.g., blast-furnace slags from steel production, fly ashes from coal-fired power plants, and waste glasses) or naturally derived materials (e.g., volcanic ashes and calcined clays) can be used to partially replace Portland cement (PC) [29][30][31] to lower the CO 2 emissions associated with the use of PC (the dominant cement product in the current market) in concrete.…”

Section: Introductionmentioning

confidence: 99%

“…The density data used to train ML models in the aforementioned studies is mostly experimentally determined, whereas a recent study 24 employed high‐throughput atomistic simulations to estimate the density and elastic moduli of silicate‐based glasses, which were then used to train the ML model (i.e., least absolute shrinkage and selection operator with a gradient boost machine). In addition to density, ML techniques have also been increasingly used to predict other glass properties, including elastic properties, glass transition temperature, coefficient of thermal expansion, liquidus temperature, refractive index, and chemical durability 25–28 …”

Section: Introductionmentioning

confidence: 99%

Data‐driven prediction of room‐temperature density for multicomponent silicate‐based glasses

Gong

Olivetti

2023

J Am Ceram Soc.

View full text Add to dashboard Cite

Density is one of the most commonly measured or estimated material properties, especially for glasses and melts that are of significant interest to many fields, including metallurgy, geology, materials science, and sustainable cements. Here, three types of machine learning models (i.e., random forest [RF], artificial neural network [ANN], and Gaussian process regression [GPR]) have been developed to predict the room‐temperature density of glasses in the compositional space of CaO–MgO–Al2O3–SiO2–TiO2–FeO–Fe2O3–Na2O–K2O–MnO (CMASTFNKM), based on ∼2100 data points mined from ∼140 literature studies. The results show that the RF, ANN, and GPR models give accurate predictions of glass density with R2 values, root mean square error (RMSE), and mean absolute percentage error (MAPE) of ∼0.95–0.97, ∼0.03–0.04 g/cm3 and ∼0.63%–0.83%, respectively, for the 15% testing set, which are more accurate compared with empirical density models based on ionic packing ratio (with R2 values, RMSE, and MAPE of ∼0.28–0.91, ∼0.05–0.15 g/cm3, and ∼1.40%–4.61%, respectively). Furthermore, glass density is shown to be a reliable reactivity indicator for a range of CaO–Al2O3–SiO2 (CAS) and volcanic glasses due to its strong correlation (R2 values above ∼0.90) with the average metal–oxygen dissociation energy (a structural descriptor) of these glasses. Analysis of the predicted density–composition relationships from these models (for selected compositional subspaces) suggests that the single‐layer ANN and GPR models exhibit a certain level of transferability (i.e., ability to extrapolate to compositional spaces not [or less] covered in the database) and capture some known features, including the mixed alkaline earth (or alkali) effects.

show abstract

Predicting glass properties by using physics- and chemistry-informed machine learning models

Cited by 12 publications

References 35 publications

Enhancing glass property predictions through ab initio‐derived descriptors

Enhancing glass property predictions through ab initio‐derived descriptors

Beyond nature's base pairs: machine learning-enabled design of DNA-stabilized silver nanoclusters

Data‐driven prediction of room‐temperature density for multicomponent silicate‐based glasses

Contact Info

Product

Resources

About