Multi-layer perceptron model for photo-voltaic material properties prediction.

data and code to reduplicate paper: Topological representations of crystalline compounds for the machine-learning prediction of materials properties

Materials representation plays a key role in machine learning based prediction of materials properties and new materials discovery. Currently both graph and 3D voxel representation methods are based on the heterogeneous elements of the crystal structures. Here, we propose to use electronic charge density (ECD) as a generic unified 3D descriptor for materials property prediction with the advantage of possessing close relation with the physical and chemical properties of materials. We developed an ECD based 3D convolutional neural networks (CNNs) for predicting elastic properties of materials, in which CNNs can learn effective hierarchical features with multiple convolving and pooling operations. Extensive benchmark experiments over 2,170 Fm-3m face-centered-cubic (FCC) materials show that our ECD based CNNs can achieve good performance for elasticity prediction. Especially, our CNN models based on the fusion of elemental Magpie features and ECD descriptors achieved the best 5-fold cross-validation performance. More importantly, we showed that our ECD based CNN models can achieve significantly better extrapolation performance when evaluated over non-redundant datasets where there are few neighbor training samples around test samples. As additional validation, we evaluated the predictive performance of our models on 329 materials of space group Fm-3m by comparing to DFT calculated values, which shows better prediction power of our model for bulk modulus than shear modulus. Due to the unified representation power of ECD, it is expected that our ECD based CNN approach can also be applied to predict other physical and chemical properties of crystalline materials.

A combination of lightweight, high specific strength, and good castability make magnesium alloys a promising engineering material for the automotive and aerospace industries. Vehicle weight reduction is one of the major means available to improve automotive fuel efficiency. High-strength steels, Aluminium (Al), and polymers are already being used to reduce weight significantly, but substantial additional reductions could be achieved by greater use of low-density magnesium (Mg) and its alloys. This project herein, therefore, relies on the use of machine learning, to assist in the development of A.I. to predict alloy compositions that are potentially useful for future metallic alloys. This study shows how a machine learning approach is able to offer acceptable precision predictions with respect to the main mechanical properties of metals.

Three machine learning models develop to predict the young's modulus of four CFRPs that are the combination of two fibers and two resins.

Machine learning pipeline for next-generation battery materials discovery. Currently focused on predicting electronic band gaps using MatBench datasets and graph neural networks (GNNs). The goal is to extend the framework toward broader materials property prediction and generative design for advanced lithium- and sodium-ion battery compounds.

This repository contains a synthetic dataset developed for the prediction of physical properties of aluminum wire rods, such as Ultimate Tensile Strength (UTS), elongation, and electrical conductivity. The dataset is designed to help machine learning practitioners and researchers in the field of materials science and manufacturing

SISSO, AUTOFEAT, Ensemble on MXenes

Using Machine Learning for Material Properties Prediction in Glass Production

To predict materials from a list of available materials which can be used for the production of solar cells based on their band gap energy and other properties.

A collection of scripts and notebooks for materials property prediction using machine learning.

A comprehensive, interactive learning path for applying machine learning to materials discovery, property prediction, and atomistic simulations.

A Machine learning model for the prediction of target properties for the evaluation of new materials as good transparent conductors

A machine learning approach to discover and evaluate novel cathode and anode materials for electric vehicle batteries using materials databases and property prediction models.

Machine learning framework for predicting bulk modulus of inorganic materials using RF, SVR, GBR, XGBoost, and CatBoost. Integrates Matminer/Pymatgen features, SHAP interpretability, and GPAW-based DFT validation, enabling accurate, scalable, and cost-effective materials property prediction.

Machine Learning Model for Material's Property Prediction

Machine learning potential energy model for materials property prediction.

CSCE 585 - Machine Learning Systems Project (Federated Learning for Materials Property Prediction)

A machine learning (ML) driven software for Property Prediction of MATerial

Machine learning projects for molecular and materials science (RDKit, cheminformatics, and property prediction).

Machine learning regression models used for model prediction of material properties from composition

Code for Paper: Advancing material property prediction: using physics-informed machine learning models for viscosity

Codes used within my research in applications of regression based machine learning for prediction of thermoelectric material properties.

Machine learning projects for materials science applications, including property prediction, feature engineering, and data-driven analysis of composition–structure–property relationships using Python and scientific computing tools.

A machine learning approach to identify and evaluate promising cathode and anode materials for electric vehicle batteries by combining materials databases, property prediction, and cost analysis.

A machine learning approach to discover and evaluate novel cathode and anode materials for high-performance electric vehicle batteries using materials databases and property prediction models.

A machine learning approach to identify and evaluate promising cathode and anode materials for electric vehicle batteries by combining materials databases, property prediction, and cost analysis.

Machine learning framework for predicting material properties and screening Thermal Protection System (TPS) candidate materials using linear regression and random forest models. The project includes ablation studies, property prediction, and ranking of high-temperature materials for aerospace applications.

A machine learning approach to identify and evaluate promising cathode and anode materials for electric vehicle batteries using multi-database integration and property prediction models.

Machine Learning-based prediction of Shear Modulus for oxide glasses using experimental composition data and regression algorithms. Includes data preprocessing, visualization, model training, cross-validation, and performance comparison for materials-informatics-driven property prediction.