Material Composition Property Prediction Using Chemical Structure Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods face difficulties in accurately predicting the physical properties of compositions obtained by combining raw materials and in identifying suitable combinations that meet desired material properties.
Innovation Solution
An information processing system utilizing machine learning models to predict physical properties based on chemical fingerprints, SMILES strings, or chemical graph structures, and allowing for genetic algorithms to optimize raw material combinations, while also providing information on suitable raw materials and manufacturers.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If machine learning models are used to predict physical properties of compositions, then prediction accuracy is improved, but computational complexity and data processing requirements increase
Solution Approach 1:
The system segments the prediction task into multiple specialized machine learning models, each trained on specific types of input data (chemical fingerprints, SMILES strings, chemical graph structures). This division allows each model to focus on specific data formats and prediction targets, improving overall accuracy while managing computational complexity through modular architecture.
Solution Approach 2:
The system introduces intermediary data representation layers that convert diverse chemical input formats (SMILES, fingerprints, graph structures) into standardized numerical representations that the machine learning models can process efficiently. These intermediary representations bridge the gap between complex chemical data and model inputs, reducing computational burden while preserving prediction accuracy.
2Reliability
If multiple types of chemical data (fingerprints, SMILES, graph structures) are input to improve prediction accuracy, then the comprehensiveness of input information increases, but data processing time and computational resources increase
Solution Approach 1:
The system performs preliminary conversion and standardization of diverse chemical data formats into unified numerical representations before feeding them into the prediction models. By pre-processing and normalizing the input data, the system reduces the computational burden during actual prediction, decreasing processing time while maintaining comprehensive input information.
Solution Approach 2:
The system transforms different types of chemical data (chemical fingerprints, SMILES strings, graph structures) into standardized numerical parameter representations that can be efficiently processed by machine learning models. This parameter transformation maintains the comprehensive chemical information while enabling faster computational processing through uniform data structures.
3Adaptability or versatility
If genetic algorithms are used to optimize raw material combinations, then the ability to identify suitable combinations meeting desired properties is improved, but computational time and algorithm complexity increase
Solution Approach 1:
The genetic algorithm implementation incorporates feedback mechanisms where prediction results from machine learning models are used to evaluate and rank candidate raw material combinations. This feedback loop allows the algorithm to efficiently converge on optimal combinations by learning from previous evaluations, reducing the number of iterations needed and thereby decreasing total computational time while maintaining high adaptability.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Comprising at least one processor obtaining a combination of information identifying each of the raw materials received from the user and the amount of each of the raw materials, and obtaining a predicted value of a physical property of the property name to be predicted for a composition comprising each of the raw materials by inputting into a first machine learning model at least one of the chemical fingerprints, SMILES strings or chemical graph structure data or product name or substance name corresponding to each of the raw materials and the amount of each of the raw materials, or by inputting into a second machine learning model a set of values based on at least one of the chemical fingerprints, SMILES strings or chemical graph structure data or product name or substance name corresponding to each of the raw materials and the amount of each of the raw materials, wherein the first machine learning model is a model in which parameters are adjusted so that it can predict outputs from inputs by means of a learning data set that takes as inputs at least one of the chemical fingerprints, SMILES strings or chemical graph structure data or product names or substance names corresponding to each of the raw materials and the amount of each of the said raw materials, and as takes as outputs physical property values of the target property names to be predicted, and the second machine learning model is a model in which parameters are adjusted so that it can predict outputs from inputs by means of a learning data set that takes as inputs a set of values based on at least one of the chemical fingerprints, SMILES strings or chemical graph structure data or product names or substance names corresponding to each of the raw materials and the amount of each of the raw materials, and takes as outputs physical property values of the target property names to be predicted.