Scholarlink

The integration of machine learning technologies, especially convolutional neural networks (CNN) and Long Short Term Memory (LSTM) in the extraction of material properties from the periodic table represents a significant advance in materials science research. CNN are widely used in artificial intelligence and machine learning applications due to their ability to handle structured complex data, including text, images, and even multi-dimensional data. The aim of this study is to understand the relationship between the energy of formation of compounds with the chemical formula ABX3 referred to as Perovskites and the arrangement in the periodic table of the elements that form them, using CNN-LSTM hybrid model. The model is trained on the Open Quantum Materials Database (OQMD) dataset and achieves a high accuracy of 0.98 and a mean prediction error of 0.055 compared to Density Functional Theory (DFT) calculations. These results demonstrate the ability of CNN and LSTM to integrate structural information from the periodic table to predict the properties of materials, paving the way for the development of new technologies to design materials and improve energy efficiency in the future.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Material Property Prediction with CNN-LSTM Hybrid Models and Periodic Table as Input Representation

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Saad Boudabia,
  • Youcef Mahdjoub,
  • Ahmed Draoui,
  • Fatima Zohra Bensalem,
  • Kheira Laiourate

摘要

The integration of machine learning technologies, especially convolutional neural networks (CNN) and Long Short Term Memory (LSTM) in the extraction of material properties from the periodic table represents a significant advance in materials science research. CNN are widely used in artificial intelligence and machine learning applications due to their ability to handle structured complex data, including text, images, and even multi-dimensional data. The aim of this study is to understand the relationship between the energy of formation of compounds with the chemical formula ABX3 referred to as Perovskites and the arrangement in the periodic table of the elements that form them, using CNN-LSTM hybrid model. The model is trained on the Open Quantum Materials Database (OQMD) dataset and achieves a high accuracy of 0.98 and a mean prediction error of 0.055 compared to Density Functional Theory (DFT) calculations. These results demonstrate the ability of CNN and LSTM to integrate structural information from the periodic table to predict the properties of materials, paving the way for the development of new technologies to design materials and improve energy efficiency in the future.