Development of a machine learning computational technique for estimation of molecular diffusivity of nonelectrolyte organic molecules in aqueous media

• Published in: Journal of molecular liquids Vol. 353; p. 118763
• Main Authors: Hagos Aregawi, Beyene, Diana, Tazeddinova, Su, Chia-Hung, El-Shafay, A.S., Alashwal, May, Felemban, Bassem F., Zwawi, Mohammed, Algarni, Mohammed, Wang, Fu-Ming
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 01-05-2022
• Subjects: Artificial intelligence; Computational modeling; Molecular diffusivity; SVM; Molecular diffusivity; SVM; Computational modeling; Artificial intelligence
• ISSN: 0167-7322; 1873-3166
• DOI: 10.1016/j.molliq.2022.118763

•Development of a machine learning method for simulation of diffusivity•Fine Tree and SVM study on prediction of molecular diffusivity•Validation and training the neural model using the measured data
A model was developed based on machine learning technique to predict the molecular diffusivity of organic compounds in water at infinite dilution. The considered organic compounds are nonelectrolyte and diverse to provide a comprehensive method for prediction of diffusivity at infinite dilution and temperature of 25 °C. Two different machine learning techniques including Tree fine and Fine Gaussian SVM are utilized in this work for estimation of molecular diffusivity of organic molecules into aqueous media. The inputs parameters were taken as the functional groups of the molecule which is equal to 148 groups. To train the employed machine learning algorithms, 3000 datasets are randomly chosen and then randomized again using the algorithms. The results of simulations indicated that the Fine Tree model outperformed the SVM method with great accuracy and high R coefficients in estimation of diffusion coefficients.