Predicting Single-Substance Phase Diagrams: A Kernel Approach on Graph Representations of Molecules

Xiang, Yan; Tang, Yuhang; Liu, Hongyi; Lin, Guang; Sun, Huai

doi:10.1021/acs.jpca.1c02391

Cited by 9 publications

(24 citation statements)

References 52 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For more detailed information about MGK, we refer a reader to references. 18,[28][29][30] Gaussian processes are used for regression and classification tasks. 32…”

Section: Marginalized Graph Kernel Methodsmentioning

confidence: 99%

“…29 Xiang et al developed normalized marginalized graph kernels (nMGK) for predictions of thermodynamic properties of pure organic liquids. 30 Naturally, it is interesting to compare the two different Weisfeiler-Lehman approaches, i.e., MPNNs and MGK, to understand their advantages and disadvantages. In this work, we evaluate the performance of MGK coupled with Gaussian process regression and classification (GP-MGK) using the data sets commonly used in benchmark studies.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

A Comparative Study of Marginalized Graph Kernel and Message-Passing Neural Network

Xiang

Tang

Lin

et al. 2021

J. Chem. Inf. Model.

Self Cite

View full text Add to dashboard Cite

This work proposes a state-of-the-art hybrid kernel to calculate molecular similarity. Combining with Gaussian process models, the performance of the hybrid kernel in predicting molecular properties is comparable to that of the Directed Message Passing Neural Network (D-MPNN). The hybrid kernel consists of a marginalized graph kernel (MGK) and a radial basis function (RBF) kernel that operates on molecular graphs and global molecular features, respectively. Bayesian optimization was used to get the optimal hyperparameters for both models. The comparisons are performed on 11 publicly available data sets. Our results show that the predictions of both models are correlated with similar performance, and the ensemble prediction of both models performs better than either of them. Through principal component analysis, we found that the features extracted by the hybrid kernel are similar to those extracted by D-MPNN. The advantage of D-MPNN lies in computational efficiency, while the advantage of the graph kernel models lies in the inherent uncertainty quantification and accurate uncertainty quantification.

show abstract

“…For more detailed information about MGK, we refer a reader to references. 18,[28][29][30] Gaussian processes are used for regression and classification tasks. 32…”

Section: Marginalized Graph Kernel Methodsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

A Comparative Study of Marginalized Graph Kernel and Message-Passing Neural Network

Xiang

Tang

Lin

et al. 2021

J. Chem. Inf. Model.

Self Cite

View full text Add to dashboard Cite

show abstract

“…28 More details can be found in references. 15,[25][26][27]…”

Section: Normalized Marginalized Graph Kernel Methodsmentioning

confidence: 99%

“…Therefore, an efficient algorithm to find the optimal hyperparameters of the graph kernel is needed. In the current situation, the advantage of GNNs is that the calculation is more efficient, while the advantage of the graph core model is uncertainty qualification 27 and active learning. 26…”

Section: Uncertainty Analysis Of Gpr-mgkmentioning

confidence: 99%

“…26 Xiang et al developed normalized marginalized graph kernels (nMGK) for molecules and constructed accurate prediction models for various thermodynamic and transport properties of pure substances. 27 In this paper, we aim to benchmark the marginalized graph kernels using the direct message passing neural network (D-MPNN) 13 as a strong baseline.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

A Comparative Study of Marginalized Graph Kernel and Message Passing Neural Network

Xiang

Tang²,

Lin³

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

<p>This work presents a state-of-the-art hybrid kernel for molecular property predictions. The hybrid kernel consists of a marginalized graph kernel that operates on molecular graphs and radial basis function kernels that operate on global molecular features. Direct message passing neural network (D-MPNN) with global molecular features is used as strong baselines. After using Bayesian optimization to find the optimal hyperparameters, we benchmark the models on 11 publicly available data sets. Our results show that the prediction of the graph kernel is correlated to the prediction of D-MPNN, which indicates that the molecular representation learned from D-MPNN is very close to the reproducing kernel Hilbert space generated by the hybrid kernel. These results may provide clues for research on the interpretability of graph neural networks. In addition, ensembling the graph kernel models with D-MPNN is the best. The advantage of D-MPNN lies in computational efficiency, and the advantage of the graph kernel model lies in the inherent uncertainty qualification of Gaussian process regression. All codes for graph kernel machines used in this work can be found at <a href="https://github.com/Xiangyan93/Chem-Graph-Kernel-Machine">https://github.com/Xiangyan93/Chem-Graph-Kernel-Machine</a>.</p>

show abstract

Prediction of Phase Diagrams and Associated Phase Structural Properties

Zipoli

Viterbo

Schilter

et al. 2022

Ind. Eng. Chem. Res.

View full text Add to dashboard Cite

The knowledge of phase diagrams is crucial to understand materials and to design new ones with better properties. However, elucidating phase diagrams is a difficult task, both experimentally and theoretically. In this work, we address the problem of predicting phase diagrams and crystal structures using a data-driven approach. We used machine learning methods to predict the stable phases using composition, temperature, and pressure extracted from a set of diagrams contained in the NIST database. We used the extracted data to train machine learning algorithms to predict the stable phases and the chemical formula of the stable compounds, including structural features such as Bravais lattices, crystal types, and the local atomic environments of the inorganic compounds contained in the ICSD database. The computationally efficient data-driven methods presented in this paper will aid material scientists in estimating the structure of virtually any mixture of elements in any proportion over a wide temperature range.

show abstract

Predicting Single-Substance Phase Diagrams: A Kernel Approach on Graph Representations of Molecules

Cited by 9 publications

References 52 publications

A Comparative Study of Marginalized Graph Kernel and Message-Passing Neural Network

A Comparative Study of Marginalized Graph Kernel and Message-Passing Neural Network

A Comparative Study of Marginalized Graph Kernel and Message Passing Neural Network

Prediction of Phase Diagrams and Associated Phase Structural Properties

Contact Info

Product

Resources

About