Search

This study explored the potential of building an image-based quality control system for particleboard manufacturing. Single-layer particleboards were manufactured under 27 operating conditions and their modulus of elasticity (MOE) and the modulus of rupture (MOR) were determined. Subsequently, images of the upper surface, lower surface, and cross-section of each specimen were collected. Two types of convolutional neural networks (CNNs) were designed: a single-input CNN processing one image and a multi-input CNN capable of analyzing multiple images simultaneously. Their prediction accuracies were then compared. Among the single-input CNNs, the cross-sectional image yielded the best prediction accuracy for both the MOE and MOR. For multi-input CNNs, the combination of the upper surface and cross-sectional images produced the highest scores when the model merged the information from each image at early stage, outperforming single-input CNNs. Adding density information to multi-input CNNs significantly improved prediction accuracy for both MOE and MOR, achieving optimal results. Regression activation maps were constructed to visualize the image features that were strongly correlated with the predicted results. For MOE prediction, the precise location of phenol formaldehyde (PF) resin and particle alignment were crucial. For MOR prediction, the interface between particles and PF resin was the key. [ABSTRACT FROM AUTHOR]

Recent studies with the new program RoseBoom© claim it can predict reliable detonation parameters only based on the structural formula, without the need of a heat of formation or density obtained using a different method. In this study, it was investigated how big the impact on the calculated detonation parameters is, when one uses the density and heat of formation predicted by RoseBoom2.2© vs. densities and the heat of formations published with the corresponding molecules. A range of traditionally used models in terms of the sensitivity to the accuracy to the input values is tested. Furthermore, it proofs the need to agree on one software for predicting the performance of energetic materials, starting with the input of values of energetic materials. Additionally, it puts further trust into the predictions by RoseBoom© and raises awareness of the uncertainty of published performance values. [ABSTRACT FROM AUTHOR]

Artificial Intelligence (AI) is pivotal in advancing science, including nanomaterial studies. This review explores AI‐based image processing in nanoscience, focusing on algorithms to enhance characterization results from instruments like scanning electron microscopy, transmission electron microscopy, X‐ray diffraction, atomic force microscopy etc. It addresses the significance of AI in nanoscience, challenges in advancing AI‐based image processing for nano material characterization, and AI's role in structural analysis, property prediction, deriving structure‐property relations, dataset augmentation, and improving model robustness. Key AI techniques such as Graph Neural Networks, adversarial training, transfer learning, generative models, attention mechanisms, and federated learning are highlighted for their contributions to nano science studies. The review concludes by outlining persisting challenges and thrust areas for future research, aiming to propel nanoscience with AI. This comprehensive analysis underscores the importance of AI‐powered image processing in nanomaterial characterization, offering valuable insights for researchers. [ABSTRACT FROM AUTHOR]

Machine learning (ML) has proven to be a powerful tool, significantly speeding up and simplifying the development of new materials while enhancing their functional characteristics. In recent years, there has been an exponential growth in the number of scientific publications exploring the use of ML in materials science. Using this approach, various materials, including magnetic ones, are being actively developed and studied. This article aims to critically review research that applies ML to predict the functional characteristics of soft and hard magnetic materials. The paper is divided into three sections: the first outlines the basic principles and algorithms of machine learning, highlighting its use in addressing practical materials science challenges; the second discusses recent advances in developing magnetic functional alloys using ML; the last section provides a critical analysis of the use of machine learning methods in this area, analyzes its advantages and disadvantages, and gives recommendations for organizing such research. [ABSTRACT FROM AUTHOR]

Increasing the diversity of bio‐based polymers is needed to address the combined problems of plastic pollution and greenhouse gas emissions. The magnitude of the problems necessitates rapid discovery of new materials; however, identification of appropriate chemistries maybe slow using current iterative methods. Machine learning (ML) methods could significantly expedite new material discovery and property identification. Here, PolyAGM, a ML algorithm using graph kernel methods, is introduced and used to predict the properties of block copolymers and identify the responsible structural ‘motifs’. It applies a “fingerprinting” method to convert Graph representations of polymers into numerical vectors. The Graphs explicitly encode the entire copolymer of atoms and bonds such that the sequencing of chemical features and polymer chain length are included, alongside relevant stereochemical information. PolyAGM gives predictions for both thermal and mechanical properties that are in good agreement with experimental measurements. This work focuses on predicting the properties of bio‐derived ABA‐block polymer thermoplastic elastomers, but the general fingerprinting technique of PolyAGM should be relevant to other application fields. [ABSTRACT FROM AUTHOR]

Accurate physical property prediction of newly developed compounds is vital across various industrial sectors, particularly for the customization of fuels and additives. Artificial intelligence (AI) has recently emerged as a best practice in numerous industrial fields because of its capacity for swift and precise calculations. While conventional methods such as group contribution models have been used to estimate physical properties from molecular structure, AI offers significant potential for improving the predictive accuracy. Thus, this work focuses on developing an AI model to predict key properties - boiling points, melting points, and flashpoints - of various hydrocarbons, to demonstrate the AI's superior predictive capabilities. A dataset consisting of 202 organic compounds was created and multilayer perceptron (MLP) neural networks were employed to estimate these properties using atomic numbers, functional groups, and molecular complexity as inputs. The model's performance was evaluated and compared against conventional group contribution methods on the same dataset. The AI model was further tested on new acetal compounds, revealing its broader applicability in both fuel and chemical sectors. Results show that the AI outperformed conventional methods, excelling in 5 out of 8 hydrocarbon types for boiling points, 7 for melting points, and all 8 for flashpoints. [ABSTRACT FROM AUTHOR]

Abstract—An analysis of current trends and opportunities for the application of artificial intelligence (AI) in materials science and concrete technology, including 3D printing in construction, is presented. The key role of AI in predicting material properties, developing new materials, and quality control is highlighted. By analyzing large volumes of data collected from numerous studies, AI can suggest optimal parameters to achieve desired material properties, thereby reducing costs and increasing production efficiency. Existing rheological models, such as the Bingham–Shvedov model or the Herschel–Bulkley model, describe material behavior based on specific equations and parameters. These models can be useful in predicting concrete properties, especially when data on its component composition is available. However, these models may be limited in their predictive accuracy, particularly for nonstandard or novel materials. It has been found that machine learning and neural networks have the potential to provide accurate predictions of rheological and physicomechanical properties of concrete materials, considering multiple parameters that influence material characteristics, including chemical and mineralogical composition, as well as structural features. The combination of experimental data and AI can successfully optimize compositions and properties during production, reducing costs and research/testing time, and opening new opportunities for researchers and engineers in the field of materials science. Machine-learning algorithms such as XGBoost, LightGBM, Catboost, and NGBoost demonstrate high predictive accuracy and have become powerful tools in the design of concrete compositions and innovative technologies. The analysis of Shapley additive explanations allows us to understand which parameters of a concrete mixture have the greatest influence on its characteristics. [ABSTRACT FROM AUTHOR]

High costs and time consuming associated with experimental trial-and-error result in low efficiency, creating an urgent need for a more effective strategy for ultra-high temperature ceramic matrix composites (UHTCMCs) development. Inspired by the exceptional performance of machine learning (ML) algorithms across various domains, this work employs ML algorithms to construct models and conduct in-depth analysis of the key factors and their patterns influencing the ablation resistance of UHTCMCs. A set of 26 dimensional features that could potentially impact the ablation resistance of UHTCMCs were established based on domain knowledge. Eight typical ML models were used to build and predict the linear ablation rate (LAR) of UHTCMCs. Results show that the random forest model has optimal generalization performance, with mean absolute error (MAE), mean squared error (MSE), and coefficient of determination (R 2) being 2.75 μm s−1 and 7.3 (μm s−1)2, and 0.71 respectively. The Shapley additive explanations values based on the random forest model reveal that the key features affecting the LAR of UHTCMCs are ranked as average melting point of ceramics (AMPC) > thermal conductivity of material (TCM) > thermal expansion coefficient of oxides (TECO) > fabrication temperature of material (FTM), all showing a negative influence on the LAR. Symbolic regression further indicates that AMPC, TCM, and TECO have an exponential negative correlation with LAR. These data-driven conclusions have been thoroughly validated through the use of C f /(TiZrHfNbTa)C composites. The established model can accelerate the discovery of material knowledge and provide reliable guidance for UHTCMC development. [ABSTRACT FROM AUTHOR]

Microstructural modeling at progressive length scales can enable the prediction of thermal and mechanical properties of thermal sprayed coatings with hierarchical features. Object-oriented finite (OOF2) element modeling conducted using microstructural images, although a powerful technique, has been employed to a limited extent in thermally sprayed materials. Consequently, there is little scientific understanding of the efficiency of the OOF2 technique for estimating bulk properties. For the first time, this study provides a comprehensive analysis of these factors' role in the OOF2 technique's capability to predict thermal and mechanical properties in ceramic and metallic coatings manufactured by plasma spray, high-velocity oxyfuel (HVOF) spray, wire_arc spray, and cold spray. The prediction efficiency generally increases for larger grain sizes as overall microstructural features are captured even at lower magnifications. The same effect is obtained in microstructures having lower and uniformly shaped pores. The data on the porosity suggest that OOF2 predictions are most accurate when conducted on coatings manufactured using sintered feedstock because of the dense powder. In contrast, OOF2 predictions are the least accurate when hollow spherical (HOSP) feedstock having empty cores is used. These multiscale facets of microstructure, porosity, etc., thus, highlight the importance of the selection of the representative volume element for accurate analysis in OOF2, which, depending upon the process, is captured at 300× − 500× for HVOF and wire-arc spray, and 1000× − 15,000× magnifications for plasma and cold spray. This overall assessment charts the relative importance of variables such as grain size, porosity, and feedstock as compared to that of the process and anisotropy in the prediction of properties in thermally sprayed coatings. While these conclusions are based on the limited literature of 37 articles, this study makes a bold attempt towards a guidebook for future thermal spray researchers in conducting more accurate OOF2 analysis. [ABSTRACT FROM AUTHOR]

Molecular graph representation learning has been widely applied in various domains such as drug design. It leverages deep learning techniques to transform molecular graphs into numerical vectors. Graph Transformer architecture is commonly used for molecular graph representation learning. Nevertheless, existing methods based on the Graph Transformer fail to fully exploit the topological structural information of the molecular graphs, leading to information loss for molecular representation. To solve this problem, we propose a novel molecular graph representation learning method called MTS-Net (Molecular Topological Structure-Network), which combines both global and local topological structure of a molecule. In global topological representation, the molecule graph is first transformed into a tree structure and then encoded by employing a hash algorithm for tree. In local topological representation, paths between atom pairs are transcoded and incorporated into the calculation of the Transformer attention coefficients. Moreover, MTS-Net has intuitive interpretability for identifying key structures within molecules. Experiments on eight molecular property prediction datasets show that MTS-Net achieves optimal results in three out of five classification tasks, the average accuracy is 0.85, and all three regression tasks. [ABSTRACT FROM AUTHOR]

A pivotal challenge in advancing inorganics optoelectronic technologies, is the precise characterization of materials' electronic attributes, with the bandgap being a critical property. Conventional approaches, heavily reliant on time‐intensive and financially demanding experimental and computational methods, such as density functional theory (DFT) calculations, face limitations due to inherent estimation errors. Machine learning methodologies are developed for the prediction of bandgaps of inorganic semiconductors but most of them are employed for datasets created by DFT calculations, hence limiting their performance. Addressing this, the study leverages machine learning methodologies, harnessing both compositional and structural features, to predict the band gaps of inorganic semiconductors with enhanced accuracy. This advancement is reinforced by the employment of an experimental bandgap dataset, which, when integrated with structural descriptors obtained from the Materials Project, significantly improves prediction capabilities. This is evidenced by the model's exceptional performance across two distinct benchmark datasets. Furthermore, the model's adeptness in predicting formation energies underscores its versatility and applicability to a broad spectrum of electronic properties. These findings suggest that the predictive accuracy of this model can be further augmented through the inclusion of additional experimental bandgap measurements and the refinement of structural descriptors. This approach offers a promising and efficient alternative to traditional methodologies, potentially accelerating the development of optoelectronic technologies. [ABSTRACT FROM AUTHOR]

Selecting suitable biodiesel for the intended application is challenging due to the significant variations in the feedstock for producing biodiesel. The available models to predict biodiesel properties have limited applicability and reliability. The present work addresses these two challenges by developing reliable models based on machine learning algorithms for predicting engine fuel properties of biodiesel and optimizing biodiesel composition for better fuel properties. The models are developed using multilinear regression (MLR), artificial neural networks (ANN), support vector machine regression with grid search (SVMGS), Bayesian optimization (SVMBO) and grey-wolf optimization (SVMGWO) for hyperparameter tuning, Gaussian process regression (GPR), random forest (RF), and adaptive neuro-fuzzy inference system (ANFIS) algorithms. The models are trained to predict viscosity, cetane number, and calorific value from 13 methyl ester constituents of 70 biodiesels. SVMGS models predicted the viscosity, cetane number, and calorific value of 33 validation samples with a mean absolute percentage error of 1.54%, 1%, and 0.43%. Biodiesel composition was optimized to minimize viscosity and maximize cetane number and calorific value. The optimized composition exhibits 3.72 cSt viscosity, 57 cetane number, and 43 MJ/kg calorific value, which can be prepared by blending 68% ± 1% camelina and 32% ± 1% coconut oil. Applying machine learning algorithms to predict biodiesel properties yielded more accurate predictions than available models. It helped find the optimal composition for improved engine characteristics. [ABSTRACT FROM AUTHOR]

Additive manufacturing (AM) is a revolutionary technology that enables production of intricate structures while minimizing material waste. However, its full potential has yet to be realized due to technical challenges such as the dependence of part quality on numerous process parameters, the vast number of design options, and the occurrence of defects. These complications may be magnified by the use of polymers and polymer composites due to their complex molecular structures, batch‐to‐batch variations, and changes in final part properties caused by small alterations in process settings and environmental conditions. Machine learning (ML), a branch of artificial intelligence, offers approaches to tackle these challenges and significantly reduce the experimental and computational time and expense. This review provides a comprehensive analysis of existing research on integrating ML techniques into polymer AM. It highlights the challenges involved in adopting ML in polymer AM, proposes potential solutions, and identifies areas for future research. [ABSTRACT FROM AUTHOR]

Language Models (LMs) have recently achieved remarkable success in natural language processing and other Artificial Intelligence (AI) applications. In this work, we adopt a language‐like representation of organic molecules and utilize LMs to address two typical tasks in the discovery of Organic Light‐Emitting Diode (OLED) materials: property prediction and structure generation. In the prediction task, the LM serves as a surrogate model of the quantum chemistry simulator for electronic properties prediction. In the generation task, the LM acts as a conditional generator for generating novel molecules with desired properties. This work demonstrates the great potential of LMs in unifying multiple tasks in OLED materials discovery within a simple but efficient framework. [ABSTRACT FROM AUTHOR]

Laser powder bed fusion (LPBF) is one of the common metal additive manufacturing technologies, which has been increasingly applied across various industries, including healthcare, manufacturing, and aerospace, owing to its advantages in customization and faster prototyping. However, acquiring accurate product properties necessitates repetitive and time-consuming measurements, which risk damaging the product. Thus, there is a pressing need to develop an automated method for predicting product properties. In this study, to forecast these properties, we documented details related to metal additive manufacturing products, encompassing both the process parameters and textural features. These features were extracted from layer-by-layer images using the gray-level co-occurrence matrix (GLCM). Subsequently, we employed machine learning (ML) models, such as support vector regression (SVR), XGBoost, and LightGBM, to predict product properties and compare their performance. The experimental results reveal stronger correlations between process parameters and texture features of three-dimensional co-occurrence matrices of the product images, compared to two-dimensional ones. Additionally, the models exhibit high predictive accuracy, especially XGBoost, and LightGBM, with R2 scores approaching 0.9 for all properties. These findings highlight the superiority and feasibility of the proposed approach. Moreover, this proposed approach holds promise in accurately predicting diverse product properties, meeting the demands of multiple application contexts. [ABSTRACT FROM AUTHOR]

The physical and mechanical properties of heat-treated wood are essential factors in assessing its appropriateness for different applications. While back-propagation (BP) neural networks are widely used for predicting wood properties, their accuracy often falls short of expectations. This paper introduces an improved Beluga Whale Optimisation (IBWO)-BP model as a solution to this challenge. We improved the standard Beluga Whale Optimisation (BWO) algorithm in three ways: (1) use Bernoulli chaos mapping to explore the entire search space during population initialization; (2) incorporate the position update formula of the Firefly Algorithm (FA) to improve the position update strategy and convergence speed; (3) apply the opposition-based learning based on the lens imaging (lensOBL) mechanism to the optimal individual, which prevents the algorithm from getting stuck in local optima during each iteration. Subsequently, we adjusted the weights and thresholds of the BP model, deploying the IBWO approach. Ultimately, we employ the IBWO-BP model to predict the swelling and shrinkage ratio of air-dry volume, as well as the modulus of elasticity (MOE) and bending strength (MOR) of heat-treated wood. The benefit of IBWO is demonstrated through comparison with other meta-heuristic algorithms (MHAs). When compared to earlier prediction models, the results revealed that the mean square error (MSE) decreased by 39.7%, the root mean square error (RMSE) by 22.4%, the mean absolute percentage error (MAPE) by 9.8%, the mean absolute error (MAE) by 31.5%, and the standard deviation (STD) by 18.9%. Therefore, this model has excellent generalisation ability and relatively good prediction accuracy. [ABSTRACT FROM AUTHOR]

Exploring the application of machine learning (ML) in energetic materials (EMs) has been a hot research topic. Accordingly, the prediction of the detonation properties of EMs using ML methods has attracted much attention. However, the predictive models for the thermal decomposition temperatures (Td) of EMs have been scarcely reported. Furthermore, the small datasets used in these reports lead to a weak generalization ability of the predictive models. This study created a dataset containing 1022 energetic molecules with Td values of 38–425 ​°C and determined an optimal predictive model through training. The gradient boost machine for regression (GBR) model yielded a coefficient of determination (R2) of 0.65 and a mean absolute error (MAE) of 27.7 for the test set. This study further explored critical features, determining that the prediction accuracy of the models was significantly influenced by descriptors representing molecular bond stability (i.e., the BCUT metrics) and atomic composition (i.e., the Molecular ID). Finally, the analysis of the outlier structure indicated that the model accuracy can be further improved by incorporating features related to molecular interactions. The results of this study help gain a deep understanding of the application of ML in the prediction of EM properties, particularly in dataset construction and feature selection.

Aim: Solubility prediction is an essential factor in rational drug design and many models have been developed with machine learning (ML) methods to enhance the predictive ability. However, most of the ML models are hard to interpret which limits the insights they can give in the lead optimization process. Here, an approach to construct and interpret solubility models with a combination of physicochemical properties and ML algorithms is presented. Methods: The models were trained, optimized, and tested in a dataset containing 12,983 compounds from two public datasets and further evaluated in two external test sets. More importantly, the SHapley Additive exPlanations (SHAP) and heat map coloring approaches were used to explain the predictive models and assess their suitability to guide compound optimization. Results: Among the different ML methods, random forest (RF) models obtain the best performance in the different test sets. From the interpretability perspective, fragment-based coloring offers a more robust interpretation than atom-based coloring and that normalizing the values further improves it. Conclusions: Overall, for certain applications simple ML algorithms such as RF work well and can outperform more complex methods and that combining them with fragment-coloring can offer guidance for chemists to modify the structure with a desired property. This interpretation strategy is publicly available at https://github.com/Pharmacelera/predictive-model-coloring and could be further applied in other property predictions to improve the interpretability of ML models.

Developing new pharmaceutical compounds is a lengthy, costly, and intensive process. In recent years, the development of Artificial Intelligence (AI), Machine Learning (ML), and Deep Learning (DL) models has drawn considerable interest in drug discovery. In this review, we discuss recent advances in the field and show how these methods can be leveraged to assist each stage of the drug discovery process. After discussing recent technical progress in the encoding of chemical information via fingerprinting and the emergence of graph-based and generative models, we examine all types of interactions, including drug-target interactions, protein-protein interactions, protein-peptide interactions, and nucleic acid-based interactions. Furthermore, we discuss recent advances enabled by DL models for the prediction of ADMET (Absorption, Distribution, Metabolism, Elimination, Toxicity) properties and of solubility. We also review applications that have emerged in the past two years with the development of models, for instance, on SARS-CoV-2 inhibitors and highlight outstanding challenges.

Abstract In recent years, interdisciplinary research has become increasingly popular within the scientific community. The fields of materials science and chemistry have also gradually begun to apply the machine learning technology developed by scientists from computer science. Graph neural networks (GNNs) are new machine learning models with powerful feature extraction, relationship inference, and compositional generalization capabilities. These advantages drive researchers to design computational models to accelerate material property prediction and new materials design, dramatically reducing the cost of traditional experimental methods. This review focuses on the principles and applications of the GNNs. The basic concepts and advantages of the GNNs are first introduced and compared to the traditional machine learning and neural networks. Then, the principles and highlights of seven classic GNN models, namely crystal graph convolutional neural networks, iCGCNN, Orbital Graph Convolutional Neural Network, MatErials Graph Network, Global Attention mechanism with Graph Neural Network, Atomistic Line Graph Neural Network, and BonDNet are discussed. Their connections and differences are also summarized. Finally, insights and prospects are provided for the rapid development of GNNs in materials science at the atomic scale.

The demand for high-performance displays is driving the development of glass substrates with extremely low strain variation during manufacturing. To achieve the necessary balance of physical properties to resist strain, the traditional compositional design methods for glass substrates like trial-and-error and classical computational techniques should be optimized. As an alternative approach, machine learning (ML) algorithms have emerged for designing new glass compositions. In this work, we conduct ML research focused on the compositional design for high-performance glass substrates. We employ three ML algorithms i.e., Random Forest (RF), Classification And Regression Tree (CART) and k-Nearest Neighbors (k-NN) to predict five physical properties that are key to the performance of alkaline-free aluminosilicate glass substrates. By using only small-to-medium dataset sizes, our model reaches a high coefficient of determination of 0.9879. Furthermore, our model achieves an accurate generalized prediction of 50 experimental data and enables the prediction and design of glass substrate compositions with advanced comprehensive properties. [ABSTRACT FROM AUTHOR]

An important aspect in the development of small molecules as drugs or agrochemicals is their systemic availability after intravenous and oral administration. The prediction of the systemic availability from the chemical structure of a potential candidate is highly desirable, as it allows to focus the drug or agrochemical development on compounds with a favorable kinetic profile. However, such predictions are challenging as the availability is the result of the complex interplay between molecular properties, biology and physiology and training data is rare. In this work we improve the hybrid model developed earlier (Schneckener in J Chem Inf Model 59:4893–4905, 2019). We reduce the median fold change error for the total oral exposure from 2.85 to 2.35 and for intravenous administration from 1.95 to 1.62. This is achieved by training on a larger data set, improving the neural network architecture as well as the parametrization of mechanistic model. Further, we extend our approach to predict additional endpoints and to handle different covariates, like sex and dosage form. In contrast to a pure machine learning model, our model is able to predict new end points on which it has not been trained. We demonstrate this feature by predicting the exposure over the first 24 h, while the model has only been trained on the total exposure. [ABSTRACT FROM AUTHOR]

With the growing prominence of environmental conservation awareness, there has been a notable surge in the exploration of renewable materials, particularly in the realm of natural fiber reinforced polymer composites. This heightened focus is underscored by the recent advancements in additive manufacturing techniques dedicated to continuous natural fiber reinforced composites (CNFRCs), which have inherently opened unprecedented avenues for the holistic customization of CNFRCs with meticulously tailored properties. This work reviewed the advanced techniques for 3D printing CNFRCs and addressed their challenges and perspectives in the future. First, the 3D printing processes of CNFRCs were reviewed, and the use of reinforcement and matrix phases was classified in detail. Then, CNFRCs were discussed in terms of their mechanical performances and novel function of shape‐changing. Further, performance optimization and prediction methods of 3D printed CNFRCs were discussed. In conclusion, a perspective on future study opportunities of 3D printed CNFRCs was provided from design, manufacturing, prediction to application. [ABSTRACT FROM AUTHOR]

Metallic glasses have received a lot of interest because of their excellent mechanical, physical, and chemical qualities. For example, they have a stronger resistivity than crystalline metals composed of the same elements and a lower viscosity coefficient. However, the difficulty in creating alloy compositions has been a concern for researchers. Traditional amorphous alloy systems design approaches, such as empirical trial-and-error methods and methods based on density functional theory (DFT), have assisted researchers in exploring numerous amorphous alloy systems during the growth of materials science over the last few decades. However, with the continuous development of materials science, these methods have been difficult to meet the needs of researchers due to their long development cycles and low efficiency. Additionally, the complex and long-range disordered structure of metallic glasses makes it difficult to understand their structure and nature in a comprehensive and clear way by conventional methods. Amorphous alloy composition design and property analysis are now often conducted using machine learning techniques because of their low experimental cost, short development cycle, strong data processing capability, and high predictive performance, among other advantages. They present new approaches and chances to address significant key bottlenecks in the field of metallic glass. In this study, the main processes of machine learning model building were introduced. Subsequently, the related studies on data pre-processing, model construction, and model validation were presented. For data pre-processing, data selection, feature engineering, and advanced data balancing methods were primarily described. In the feature engineering part, the model performance with various input features was examined, and it was shown that either employing physical properties or directly using the alloy compositions as the model input might result in high performance. Four machine learning algorithms were used to generate the machine learning model: artificial neural networks (ANN), support vector machines (SVM), random forest (RF), and extreme gradient boosting (XGBoost). A comparison indicates that SVM models work best with small data sets, whereas the performance of all other models tends to get better as the amount of training data increases. Generally, the XGBoost method outperforms several other methods and is, therefore, often used in machine learning competitions. Model validation approaches: K-fold cross-validation and leave-one-out cross-validation methods were presented. A good metallic glass performance prediction method needs to perform well in both validation methods. Finally, this study provides several possible future research directions on feature engineering, dataset construction, validation, and machine learning models.

Understanding the physical properties of distillate petroleum fuels like gasoline and diesel is very critical to ensure the normal operation of internal combustion (IC) engines with regards to processes like spray atomization, heating, evaporation etc. Two of most important physical properties are density and viscosity. Many factors such as molecular structure, molecular weight, temperature etc. effect the physical properties of the fuel. The present work deals with the development of a machine learning model for predicting the density and viscosity of petroleum fuels containing oxygenated chemical classes such as alcohols, esters, ketones and aldehydes. The model was developed using the molecular structure of the compounds expressed in the form of functional groups as inputs. The density and viscosity of 164 pure compounds spanning various chemical families and 14 blends of known compositions was collected from the literature. An artificial neural network model (ANN) for predicting density and viscosity was developed using the neural network tool in Matlab. Each of the ANN model was tested against 15% of the data and the results show that the models were able to successfully predict the density and viscosity of the unseen data points to a good accuracy. A regression coefficient of 0.99 (for density) and 0.98 (for viscosity) was obtained for the test set. The developed models can be used to predict and screen the density and viscosity of real petroleum fuels containing drop in oxygenated bio-fuels.

Polymeric materials with excellent performance are the foundation for developing high-level technology and advanced manufacturing. Polymeric material genome engineering (PMGE) is becoming a vital platform for the intelligent manufacturing of polymeric materials. However, the development of PMGE is still in its infancy, and many issues remain to be addressed. In this perspective, we elaborate on the PMGE concepts, summarize the state-of-the-art research and achievements, and highlight the challenges and prospects in this field. In particular, we focus on property estimation approaches, including property proxy prediction and machine learning prediction of polymer properties. The potential engineering applications of PMGE are discussed, including the fields of advanced composites, polymeric materials for communications, and integrated circuits.

Abstract Exploring the “composition‐microstructure‐property” relationship is a long‐standing theme in materials science. However, complex interactions make this area of research challenging. Based on the image processing and machine learning techniques, this paper proposes a multimodal fusion learning framework that comprehensively considers both composition and microstructure in prediction of the ultimate tensile strength (UTS) of Al‐Si alloys. Firstly, the composition and image information are collected from the literature and supplementary experiments, followed by the image segmentation and quantitative analysis of eutectic Si images. Subsequently, the quantitative analysis results are combined with other features for three‐step feature screening, and 12 key features are obtained. Finally, four machine‐learning models (i.e., decision tree, random forest, adaptive boosting, and extreme gradient boosting [XGBoost]) are used to predict the UTS of Al‐Si alloys. The results show that the quantitative analysis method proposed in this paper is superior to Image‐Pro Plus (IPP) software in some aspects. The XGBoost model has the best prediction performance with R2 = 0.94. Furthermore, five mixed features and their critical values that significantly affect UTS are identified. Our study provides enlightenment for the prediction of UTS of Al‐Si alloys from composition and microstructure, and would be applicable to other alloys.

Machine learning techniques offer tremendous potential for optimizing resource allocation in solving real-world problems. However, the emergence of multi-fidelity data introduces new challenges. This paper offers an overview of the definition, applications, data preprocessing methodologies, and learning approaches associated with multi-fidelity data. To validate the algorithms, we examine three widely-used learning methods relevant to multi-fidelity data through the design of multi-fidelity datasets that encompass various types of noise. As we expected, employing multi-fidelity data learning methods yields better results compared to solely using high-fidelity data learning methods. Additionally, considering the inherent various types of noise within datasets, the comprehensive correction strategy proves to be the most effective. Moreover, multi-fidelity learning methods facilitate effective decision-making processes by enabling the combination of datasets from various sources. They extract knowledge from lower fidelity data, improving model accuracy compared to models solely relying on high-fidelity data. [ABSTRACT FROM AUTHOR]

A multitask deep neural network model was trained on more than 218k different glass compositions. This model, called GlassNet, can predict 85 different properties (such as optical, electrical, dielectric, mechanical, and thermal properties, as well as density, viscosity/relaxation, crystallization, surface tension, and liquidus temperature) of glasses and glass-forming liquids of different chemistries (such as oxides, chalcogenides, halides, and others). The model and the data used to train it are available in the GlassPy Python module as free and open source software for the community to use and build upon. As a proof of concept, GlassNet was used with the MYEGA viscosity equation to predict the temperature dependence of viscosity and outperformed another general purpose viscosity model available in the literature (ViscNet) on unseen data. An explainable AI algorithm (SHAP) was used to extract knowledge correlating the input (physicochemical information) and output (glass properties) of the model, providing valuable insights for glass manufacturing and design. It is hoped that GlassNet, with its free and open source nature, can be used to enable faster and better computer-aided design of new technological glasses. [ABSTRACT FROM AUTHOR]

Clarifying the relationship between compositions, heat treatment processes, and mechanical properties of carbon steel, as the basis of material design, is challengeable, while machine learning (ML) makes this complex correlation explicit. In this work, three different mechanical properties (ultimate tensile strength, yield strength, and total elongation) were predicted based on the collected quenched and tempered (Q&T) steel dataset by six ML algorithms, in which the optimal Gaussian process regression (GPR) combined with the key descriptors by feature engineering to train an optimized ML model. Such a simplified ML model shows even better prediction accuracy. In the above training process, Bayesian optimization (BO) searches the hyperparameters efficiently. The newly collected data also achieve small prediction errors, showing good generalization capacity. To maximize the application value of the current ML model, the grid prediction of composition and process, and local interpretable model-agnostic explanations (LIME) were utilized to reveal some new insights about the quenched and tempered steels, which could shed light on the ongoing new material design. Besides, the overfitting tendency of the ML model was examined to ensure the rationality of prediction, and the influence of data amount on the prediction performance was discussed.

Machine learning has exhibited high efficiency in designing concrete. However, collecting the dataset for training machine learning models is challenging. To address this challenge, this paper develops an approach to collect concrete design data automatically based on information extraction techniques. The approach enables machine learning models to automatically track, extract, and learn knowledge embedded in data from relevant publications. The approach has been incorporated into AI-assisted design of low-carbon cost-effective ultra-high-performance concrete (UHPC) via integrating the capabilities of automatically collecting and processing data, predicting UHPC properties, and optimizing UHPC properties regarding the material cost, carbon footprint, and compressive strength. A self-updating mechanism is imparted to continuously learn available data. Such a mechanism enables the self-updatable automatic discovery of low-carbon cost-effective UHPC. The results showed increasing prediction accuracy and optimization performance of the proposed approach over time when more knowledge was learned from new data, therefore accelerating the design of UHPC.

Results of statistical studies of temperature dependences for strength properties of sheets and bars made of titanium alloys of different classes after standard annealing are summarized. On the basis of these studies polynomial models for estimating typical values of the ultimate strength at temperatures of 20–600°C are substantiated. Models for predicting the tensile strength of alloys at different operating temperatures depending upon aluminum and molybdenum equivalents are proposed. Generalized strength diagrams are constructed, which allow not only prediction of ultimate strength depending upon strength or structural equivalents, but also substantiation of optimum compositions of new alloys taking account of their operating condition requirements. [ABSTRACT FROM AUTHOR]

Material innovation plays a very important role in technological progress and industrial development. Traditional experimental exploration and numerical simulation often require considerable time and resources. A new approach is urgently needed to accelerate the discovery and exploration of new materials. Machine learning can greatly reduce computational costs, shorten the development cycle, and improve computational accuracy. It has become one of the most promising research approaches in the process of novel material screening and material property prediction. In recent years, machine learning has been widely used in many fields of research, such as superconductivity, thermoelectrics, photovoltaics, catalysis, and high-entropy alloys. In this review, the basic principles of machine learning are briefly outlined. Several commonly used algorithms in machine learning models and their primary applications are then introduced. The research progress of machine learning in predicting material properties and guiding material synthesis is discussed. Finally, a future outlook on machine learning in the materials science field is presented. [ABSTRACT FROM AUTHOR]

Understanding the physical properties of distillate petroleum fuels like gasoline and diesel is very critical to ensure the normal operation of internal combustion (IC) engines with regards to processes like spray atomization, heating, evaporation etc. Two of most important physical properties are density and viscosity. Many factors such as molecular structure, molecular weight, temperature etc. effect the physical properties of the fuel. The present work deals with the development of a machine learning model for predicting the density and viscosity of petroleum fuels containing oxygenated chemical classes such as alcohols, esters, ketones and aldehydes. The model was developed using the molecular structure of the compounds expressed in the form of functional groups as inputs. The density and viscosity of 164 pure compounds spanning various chemical families and 14 blends of known compositions was collected from the literature. An artificial neural network model (ANN) for predicting density and viscosity was developed using the neural network tool in Matlab. Each of the ANN model was tested against 15% of the data and the results show that the models were able to successfully predict the density and viscosity of the unseen data points to a good accuracy. A regression coefficient of 0.99 (for density) and 0.98 (for viscosity) was obtained for the test set. The developed models can be used to predict and screen the density and viscosity of real petroleum fuels containing drop in oxygenated bio-fuels. [ABSTRACT FROM AUTHOR]

In the development and optimization of chemical processes involving the selection of organic fluids, knowledge of the physical properties of compounds is vital. In many cases, it is complex to find experimental measurements for all substances, so it becomes necessary to have a tool to predict properties based on the characteristics of the molecule. One of the most extensively used methods in the literature is the estimation by contribution of functional groups, where properties are calculated using the constituent elements of the molecule. There are several models published in the literature, but they fail to represent a wide variety of compounds with high accuracy and simultaneously maintain a low computational complexity. The aim of this work is to develop a prediction model for eight thermodynamic properties (melting temperature, boiling temperature, critical pressure, critical temperature, critical volume, enthalpy of vaporization, enthalpy of fusion, and enthalpy of gas formation) based on the group contribution methodology by implementing a multilayer perceptron. Here, 2736 substances were used to train the neural network, whose prediction capacity was compared with other reference models available in the literature. The proposed model presents errors ranging from 1% to 5% for the different properties (except for the melting point), which improves the reference models with errors in the range of 3%–30%. Nevertheless, a difficulty in the prediction of the melting point is detected, which could represent an inherent hindrance to this methodology. [ABSTRACT FROM AUTHOR]

Abstract Collecting labeled data for many important tasks in chemoinformatics is time consuming and requires expensive experiments. In recent years, machine learning has been used to learn rich representations of molecules using large scale unlabeled molecular datasets and transfer the knowledge to solve the more challenging tasks with limited datasets. Variational autoencoders are one of the tools that have been proposed to perform the transfer for both chemical property prediction and molecular generation tasks. In this work we propose a simple method to improve chemical property prediction performance of machine learning models by incorporating additional information on correlated molecular descriptors in the representations learned by variational autoencoders. We verify the method on three property prediction tasks. We explore the impact of the number of incorporated descriptors, correlation between the descriptors and the target properties, sizes of the datasets etc. Finally, we show the relation between the performance of property prediction models and the distance between property prediction dataset and the larger unlabeled dataset in the representation space.

Artificial Intelligence (AI) is becoming an emerging technique in scientific research including novel materials discovery. In this work, we present a novel graph‐based AI workflow for discovering Organic Light‐Emitting Diode (OLED) materials. This workflow contains two graph‐based AI models: a molecular structure generative model and a molecular property predictive model. The target materials here are Red‐Prime (RP) materials, which are widely used to pair with the red light emitters in OLED devices. Based on the desired properties required by our OLED devices, we apply the AI‐based workflow to perform high‐ throughput screening for RP materials. Several novel and potential RP molecules are discovered preliminarily. [ABSTRACT FROM AUTHOR]