DeePore: A deep learning workflow for rapid and comprehensive characterization of porous materials

Rabbani, Arash; Babaei, Masoud; Shams, Reza; Wang, Ying Da; Chung, Traiwit

doi:10.1016/j.advwatres.2020.103787

Cited by 69 publications

(44 citation statements)

References 89 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Finally, they tested the PiNN library on QM9 dataset containing 50,604 organic molecules and predicted properties like internal energy U 0 , and partial charges with very low MAE values [127,128]. In 2020 itself, Rabbani and coworkers published a similar deep learning workflow DeePore for characterising porous materials [129]. The model was developed on feed-forward Convolutional neural network (CNN) and uses 30 descriptors such as pore density, tortuosity, average coordination number, average pore radius, pore sphericity, etc.…”

Section: First-order Descriptor-based Modelsmentioning

confidence: 99%

Machine learning and descriptor selection for the computational discovery of metal-organic frameworks

Mukherjee

Colón

2021

Molecular Simulation

View full text Add to dashboard Cite

Metal-organic frameworks (MOFs), crystalline materials with high internal surface area and pore volume, have demonstrated great potential for many applications. In the past decade, as large number of MOFs have come into existence, there has been an effort to model them using computers. High-throughput screening techniques in tandem with molecular simulations or ab-initio calculations are being used to calculate their properties. However, the number of MOFs that can be hypothetically created are in the millions, and thoughcomputer simulations have shown remarkable accuracy, we cannot deploy them for all structures due to their high-computational cost. In this regard, machine learning (ML)-based algorithms have proven to be effective in predicting material properties and reducing the need for expensive calculations. Adopting this methodology can save time and allow researchers to explore materials in unchartered chemical space, thus ushering an era of high-throughput in-silico material design using ML. In this work, we present what is ML, its associated workflow, selecting descriptors, and how it can help build reliable models for discovering MOFs. We present somepopular and novel ones. Thereafter, we review some of the recent studies with respect to ML-based implementation for MOF discovery emphasizing descriptors selected and the workflow adopted.

show abstract

Section: First-order Descriptor-based Modelsmentioning

confidence: 99%

Machine learning and descriptor selection for the computational discovery of metal-organic frameworks

Mukherjee

Colón

2021

Molecular Simulation

View full text Add to dashboard Cite

show abstract

“…CNNs can be classified as a deep and complex version of ANNs, in which a series of matrix manipulations and multiplications transform an input tensor of data into a target tensor (Valavi et al, 2018). Considering the high potentials of CNNs in data transformation they have been frequently used for property estimation purposes in the porous material research (Alqahtani et al, 2020;Kamrava et al, 2020;Rabbani et al, 2020;Wu et al, 2019). As an example, effective diffusivity of the porous material as a macroscopic property which is an important parameter for reaction and catalysis modeling, can be obtained using CNN models (Wu et al, 2019).…”

Section: Convolutional Neural Network (Cnns)mentioning

confidence: 99%

“…Therefore, the supreme ability of CNNs to extract the rock morphological characteristics and relate to a physical property of a porous medium was corroborated, thereby a significant computational saving is envisaged. As another attempt to tackle deep characterization of porous material, a CNN-powered workflow is presented by Rabbani et al (2020) as DeePore. The model benefits from a large dataset of semi-realistic pore-scale images with 17,700 samples and 30 features which 15 of them are 1D vectors instead of scalar properties.…”

Section: Convolutional Neural Network (Cnns)mentioning

confidence: 99%

Review of Data Science Trends and Issues in Porous Media Research With a Focus on Image‐Based Techniques

Rabbani

Fernando

Shams

et al. 2021

Water Resources Research

View full text Add to dashboard Cite

show abstract

“…Nevertheless, the prediction of scaffolds’ mechanical performance directly from the CAD models, especially for cases in which FEM cannot be applied, may require advanced ML tools capable of using more complete descriptions of these complex geometries as input, and not just representative parameters. Quite recently, 2D convolutional neural networks (CNNs), with images as input, have been effectively applied to predicting multiple properties of porous materials, which constitutes a fundamental advance [ 21 ].…”

Section: Introductionmentioning

confidence: 99%

Artificial Intelligence Aided Design of Tissue Engineering Scaffolds Employing Virtual Tomography and 3D Convolutional Neural Networks

2021

View full text Add to dashboard Cite

Design requirements for different mechanical metamaterials, porous constructions and lattice structures, employed as tissue engineering scaffolds, lead to multi-objective optimizations, due to the complex mechanical features of the biological tissues and structures they should mimic. In some cases, the use of conventional design and simulation methods for designing such tissue engineering scaffolds cannot be applied because of geometrical complexity, manufacturing defects or large aspect ratios leading to numerical mismatches. Artificial intelligence (AI) in general, and machine learning (ML) methods in particular, are already finding applications in tissue engineering and they can prove transformative resources for supporting designers in the field of regenerative medicine. In this study, the use of 3D convolutional neural networks (3D CNNs), trained using digital tomographies obtained from the CAD models, is validated as a powerful resource for predicting the mechanical properties of innovative scaffolds. The presented AI-aided or ML-aided design strategy is believed as an innovative approach in area of tissue engineering scaffolds, and of mechanical metamaterials in general. This strategy may lead to several applications beyond the tissue engineering field, as we analyze in the discussion and future proposals sections of the research study.

show abstract

DeePore: A deep learning workflow for rapid and comprehensive characterization of porous materials

Cited by 69 publications

References 89 publications

Machine learning and descriptor selection for the computational discovery of metal-organic frameworks

Machine learning and descriptor selection for the computational discovery of metal-organic frameworks

Review of Data Science Trends and Issues in Porous Media Research With a Focus on Image‐Based Techniques

Artificial Intelligence Aided Design of Tissue Engineering Scaffolds Employing Virtual Tomography and 3D Convolutional Neural Networks

Contact Info

Product

Resources

About