Insights into ensemble learning-based data-driven model for safety-related property of chemical substances

Wang, Zihao; Wen, Huaqiang; Yang, Su; Shen, Weifeng; Ren, Jingzheng; Ma, Yingjie; Li, Jie

doi:10.1016/j.ces.2021.117219

Cited by 23 publications

(15 citation statements)

References 40 publications

(47 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Before training the framework, the dataset needs to be divided into two disjoint subsets: a training set for model training with multilayer stack ensembling and n ‐repeated k ‐fold bagging, and a test set for evaluating the external predictive performance of the developed model. In this work, identical to Wang et al, 3 the training and test set account for 90% and 10% of the entire FPT dataset, respectively.…”

Section: Case Studiesmentioning

confidence: 99%

“…Toward this goal, recent studies have shown increasingly remarkable interest in applying machine learning (ML) for modeling a wide variety of molecular properties, including melting points (MPs), flash points, logP (the logarithm of the octanol-water partition coefficient), toxicity, and so forth. [2][3][4][5][6][7][8][9] These applications have demonstrated high accuracies comparable to traditional modeling methods and experimental measurements at a fraction of the computational, time, and money cost. 9,10 As shown in Figure 1, when using ML for QSPR model development, every aspect of ML applications such as feature engineering, ML model selection, and model optimization, needs to be carefully configured.…”

Section: Introductionmentioning

confidence: 99%

“…3,5,[40][41][42] as comparative case studies. These research papers reported state-of-the-art ML models that were constructed and optimized by chemical engineering experts.…”

mentioning

confidence: 99%

See 2 more Smart Citations

A scalable and integrated machine learning framework for molecular properties prediction

Chen

Song

et al. 2023

AIChE Journal

View full text Add to dashboard Cite

This work introduced a scalable and integrated machine learning (ML) framework to facilitate important steps of building quantitative structure–property relationship (QSPR) models for molecular property prediction. Specifically, the molecular descriptor generation, feature engineering, ML model training, model selection and ensembling, as well as model validation and timing, are integrated into a single workflow within the proposed framework. Unlike existing modeling methods relying upon human experts that primarily focus on model/hyperparameter selection, the proposed framework succeeds by ensembling multiple models and stacking them in multiple layers. The high efficiency and effectiveness of the proposed framework are demonstrated through comparisons with literature‐reported QSPR models using identical datasets in three property modeling case studies, that is, the flash point temperature, the melting temperature, and the octanol–water partition coefficients. While requiring much less modeling time, the resultant models by the proposed framework present better predictive performance as compared with the reference models in all three case studies.

show abstract

Section: Case Studiesmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

A scalable and integrated machine learning framework for molecular properties prediction

Chen

Song

et al. 2023

AIChE Journal

View full text Add to dashboard Cite

show abstract

“…Wang et al 118 conducted a comprehensive study on the safety-related properties of organic compounds based on Stacking. Their work considers several representative ML models, including multiple linear regression, extreme learning machine, feedforward neural network, and support vector machine.…”

Section: Ensemble Learning Enhances Model Prediction Performancementioning

confidence: 99%

A Systematic Review on Intensifications of Artificial Intelligence Assisted Green Solvent Development

Wen,

Nan,

et al. 2023

Ind. Eng. Chem. Res.

Self Cite

View full text Add to dashboard Cite

Solvents are indispensable components of chemical processes, and the application of ecofriendly, safe, and efficient solvents is vital for building green chemical processes. Nowadays, new techniques have been applied in discovering and exploring green solvents, among which artificial intelligence (AI) plays an increasingly important role in predicting their physical and chemical properties. Being able to explore the chemical space of green solvents, AI can also be utilized to screen out expected solvents and inversely design new solvents. This review introduces the application of AI assisted green solvent design, focusing on intensification techniques in the processes of green solvent design and property prediction. First, the various intensification techniques of quantitative structure−property relationships (QSPR) employed in the process of solvent property prediction are summarized, including the optimization and intensification of feature extraction, ensemble learning, uncertainty analysis, and interpretability modeling. After that, the basic principles and latest theoretical advances in the application of inverse molecular design for green solvents are reviewed, including high-throughput screening, computer-aided molecular design (CAMD), and deep generative models. Finally, new ideas are proposed for the improvement of each intensification technique in order to better match the high demands of the particular application in green solvent design.

show abstract

“…Researchers have used different types of artificial intelligence algorithms to build predictive models of chemicals, including artificial neural network (ANN), light gradient boosting (LightGBM), deep neural networks (DNN), random forest (RF), extra trees (ET), multiple linear regression (MLR), partial least squared (PLS), k-nearest neighbors (k-NN), support vector machine (SVM), and Ridge regression. − For example, Ling et al used the XGBoost algorithm combined with SHapley Additive exPlanation (SHAP) to predict the viscosities of deep eutectic solvents (DESs) at different temperatures . Shen et al used different algorithms to predict properties such as the flash point temperature of substances and obtained accurate results. − However, using chemical simulation software to calculate the viscosities of polymers under different conditions poses some difficulties and challenges. The variety and variability of polymer systems hinder generalization and standardization.…”

Section: Introductionmentioning

confidence: 99%

Strategy of Coupling Artificial Intelligence with Thermodynamic Mechanism for Predicting Complex Polymer Viscosities

Wang,

Sadowski,

2024

ACS Sustainable Chem. Eng.

View full text Add to dashboard Cite

With the environmental protection requirements brought about by the large-scale application of polymers in industrial fields, understanding the viscosities of polymers is becoming increasingly important. The different arrangements and crystallinity of the polymers make their viscosities difficult to calculate. To address this challenge, new strategies based on artificial intelligence algorithms are proposed. First, the strategy trains three artificial intelligence algorithms [extreme gradient boosting (XGBoost), convolutional neural network (CNN), and multilayer perceptron (MLP)] based on molecular descriptors of the polymer molecular properties. Next, the PC-SAFT parameters are input into the XGBoost and CNN algorithms as molecular descriptors representing the thermodynamic properties of the polymer to improve the accuracy of the algorithm prediction results. Subsequently, the Molecular ACCess Systems chemical fingerprinting was combined with the XGboost algorithm and CNN algorithm to further improve the accuracy of predicting viscosities. The XGboost algorithm was identified as the best predictive algorithm for predicting the viscosities of the polymer in different states. This discovery is expected to provide effective information for screening polymers for applications in medicine and the chemical industry.

show abstract

Insights into ensemble learning-based data-driven model for safety-related property of chemical substances

Cited by 23 publications

References 40 publications

A scalable and integrated machine learning framework for molecular properties prediction

A scalable and integrated machine learning framework for molecular properties prediction

A Systematic Review on Intensifications of Artificial Intelligence Assisted Green Solvent Development

Strategy of Coupling Artificial Intelligence with Thermodynamic Mechanism for Predicting Complex Polymer Viscosities

Contact Info

Product

Resources

About