Scheme of new materials

Patent metadata

Field	Value
Publication number	US-10366779-B2
Application number	US-201514984989-A
Country	US
Kind code	B2
Filing date	Dec 30, 2015
Priority date	Dec 30, 2015
Publication date	Jul 30, 2019
Grant date	Jul 30, 2019

Abstract

A method and system are provided for predicting chemical structures. The method includes receiving, at a user interface, intended structural feature values and intended chemical property values, as vectors. The method further includes constructing, by a hardware processor, a prediction model, wherein the prediction model predicts other structural feature values from the intended structural feature values and the intended chemical property values, and automatically configuring, by the hardware processor, at least one chemical structure candidate from the other structural feature vectors. The method additionally includes evaluating the at least one chemical structure candidate to determine structural feature values and chemical property values of the at least one chemical structure candidate and performing, by the hardware processor, machine learning of a chemical structure predicting system based on the evaluated structural feature values and the evaluated chemical property values of the at least one chemical structure candidate.

First claim

The invention claimed is:

1. A method for predicting and designing chemical structures, comprising:
receiving, at a user interface, intended structural feature values in vector form;
receiving, at the user interface, intended chemical property values in vector form;
compressing, by a hardware processor, the vector form of the intended structural feature values into a scalar form of the intended structural feature values;
constructing, by the hardware processor, a prediction model, wherein the prediction model predicts other structural feature values in scalar form from the scalar form of the intended structural feature values and the vector form of the intended chemical property values;
automatically configuring, by the hardware processor, at least one chemical structure candidate from the other structural feature values in vector form;
evaluating the at least one chemical structure candidate to determine structural feature values and chemical property values of the at least one chemical structure candidate; and
designing at least one chemical structure for constructing at least one chemical material, including performing, by the hardware processor, machine learning based on the determined structural feature values and the determined chemical property values of the at least one chemical structure candidate.

2. The method of claim 1 , wherein the intended structural feature values include at least one of a number of heavy atoms, a number of ring structures, a number of non-carbon atoms, a number of constituents, a number of double bonds, and a number of triple bonds.

3. The method of claim 1 , further comprising decompressing, by the hardware processor, the scalar form of the other structural feature values into the vector form of the other structural feature values, including performing inverse dimension reduction of the other structural feature values to transform the scalar form of the other structural feature values from the scalar form to the vector form, and wherein the inverse dimension reduction comprises at least one of inverse Principal Component Analysis and a Stacked Auto Encoder.

4. The method of claim 1 , wherein the prediction model is constructed by combining a dimension reduction method and a regression method.

5. The method of claim 4 , wherein the dimension reduction method comprises at least one of Principal Component Analysis and a Stacked Auto Encoder.

6. The method of claim 1 , wherein the at least one chemical structure candidate is constructed by manipulating Simplified Molecular-Input Line-Entry System strings.

7. The method of claim 6 , further comprising visually displaying the at least one chemical structure candidate on a display device.

8. The method of claim 1 , wherein the chemical properties of the at least one chemical structure candidate are evaluated using at least one of a first principle simulation and a Molecular Dynamics method.

9. A computer program product for predicting and designing chemical structures, the computer program product comprising a non-transitory computer readable storage medium having program instructions embodied therewith, the program instructions executable by a computer to cause the computer to perform a method comprising:
receiving, at a uses interface; intended structural feature values in vector form;
receiving, at the user interface, intended chemical property values in vector form;
compressing, by a hardware processor, the vector form of the intended structural feature values into scalar forms of the intended structural feature values;
constructing, by the hardware processor, a prediction model, wherein the prediction model predicts other structural feature values in scalar form from the scalar form of the intended structural feature values and the vector form of the intended chemical property values;
automatically configuring, by the hardware processor, at least one chemical structure candidate from the other structural feature values in vector form;
evaluating the at least one chemical structure candidate to determine structural feature values and chemical property values of the at least one chemical structure candidate; and
designing at least one chemical structure for constructing at least one chemical material, including performing, by the hardware processor, machine learning based on the determined structural feature values and the determined chemical property values of the at least one chemical structure candidate.

10. The computer program product of claim 9 , wherein the intended structural feature values include at least one of a number of heavy atoms, a number of ring structures, a number of non-carbon atoms, a number of constituents, a number of double bonds, and a number of triple bonds.

11. The computer program product of claim 9 , wherein the method further comprises decompressing, by the hardware processor, the scalar form of the other structural feature values into the vector form of the other structural feature values, including performing inverse dimension reduction of the other structural feature values to transform the scalar form of the other structural feature values from the scalar form to the vector form, and wherein the inverse dimension reduction comprises at least one of inverse Principal Component Analysis and a Stacked Auto Encoder.

12. The computer program product of claim 9 , wherein the prediction model is constructed by combining a dimension reduction method and a regression method, and wherein the dimension reduction method includes at least one of Principal Component Ana and a Stacked Auto Encoder.

13. A system for predicting and designing chemical structures, comprising:
a user interface configured to receive, in vector form, intended structural feature values and intended chemical property values; and
a hardware processor configured to:
compress the vector form of the intended structural ur values into a scalar form of the intended structural feature values;
construct a prediction model, wherein the prediction model predicts other structural feature values in scalar form from the scalar form of the intended structural feature values and the vector form of the intended chemical property values;
automatically configure at least one chemical structure candidate from the other structural feature vectors in vector form;
evaluate the at least one chemical structure candidate to determine structural feature values and chemical property values of the at least one chemical structure candidate; and
design at least one chemical structure for constructing at least one chemical material by performing machine learning based on the determined structural feature values and the determined chemical property values of the at least one chemical structure candidate.

14. The system of claim 13 , herein the intended structural feature values include at least one of a number of heavy atoms, a number of ring structures, a number of non-carbon atoms, a number of constituents, a number of double bonds, and a number of triple bonds.

15. The system of claim 13 , wherein the hardware processor is further configured to decompress the scalar form of the other structural feature values into the vector form of the other structural feature values by performing inverse dimension reduction on the scalar form of the other structural feature values to transform the other structural feature values from the scalar form to the vector form, and wherein the inverse dimension reduction comprises at least one of inverse Principal Component Analysis and a Stacked Auto Encoder.

16. The system of claim 13 , wherein the hardware

Assignees

• IBM

Inventors

• Hsu Hsiang H
• Kurata Gakuto
• Masuda Koji
• Nakagawa Shigeru
• Nakamura Hajime
• Takeda Seiji

Classifications

• G16C20/70 Primary

  Machine learning, data mining or chemometrics · CPC title

• G16C20/30 Primary

  Prediction of properties of chemical compounds, compositions or mixtures · CPC title