A Cocrystal Prediction Method and Deep Learning Framework Based on Graph Neural Network

Abstract

The invention belongs to the technical field of eutectic formation prediction, and discloses a eutectic prediction method and a deep learning framework based on a graph neural network, including: eutectic sample collection; data processing; data set division; The graph neural network framework CCGNet is used for cocrystal screening, and a cocrystal prediction model is constructed under the CCGNet framework for cocrystal screening. The prediction performance of the model established by the deep learning framework CCGNet constructed by the present invention greatly surpasses the traditional machine learning model and the classical graph neural network model, and provides a high-throughput and high-accuracy solution for eutectic screening. The methodology of eutectic engineering has been developed, which is an important step towards realizing data-driven eutectic engineering design. The invention also collects a large amount of reliable cocrystal data, which provides strong data support for future cocrystal screening work based on machine learning.

Description

technical field [0001] The invention belongs to the technical field of eutectic formation prediction, and in particular relates to a graph neural network-based eutectic prediction method and a deep learning framework. Background technique [0002] At present, eutectic has become an effective way to improve the properties of materials, but only through experimental means to screen eutectic slow aging and high cost. The development of artificial intelligence technology provides another way for the screening of co-crystals, which has the advantages of rapidity and low cost. However, there are currently no large and representative co-crystal datasets, limiting the reliability of deep learning methods for application in this field. Therefore, how to use machine learning algorithms to build a reliable prediction model has become a key issue for its use due to the small sample size of the eutectic. [0003] Graph neural network is a deep learning method for graph-structured data....

AI Technical Summary

Problems solved by technology

[0005] (1) At present, only experimental methods are used to screen energetic eutectic crystals with slow aging and high cost

[0006] (2) The current lack of representative co-crystal datasets limits the reliability of machine learning methods applied in this field

[0007] The difficulty of solving the above problems and defects is: restricted by data sets and algorithms, the current literature reports that the prediction accuracy of machine learning models is low, the highest is only about 85%, and it has no practical value

Images