Methods for predicting likelihood of successful experimental synthesis of computer-generated materials by combining network analysis and machine learning

Patent metadata

Field	Value
Publication number	US-12205683-B2
Application number	US-202218147471-A
Country	US
Kind code	B2
Filing date	Dec 28, 2022
Priority date	Jun 8, 2018
Publication date	Jan 21, 2025
Grant date	Jan 21, 2025

Abstract

One aspect of the disclosure relates to systems and methods for determining probabilities of successful synthesis of materials in the real world at one or more points in time. The probabilities of successful synthesis of materials in the real world at one or more points in time can be determined by representing the materials and their pre-defined relationships respectively as nodes and edges in a network form, and computation of the parameters of the nodes in the network as input to a classification model for successful synthesis. The classification model being configured to determine probabilities of successful synthesis of materials in the real world at one or more points in time.

First claim

What is claimed is:

1. A system configured to train a classifier model to determine a probability of successful synthesis of a material at one or more points in time, the system comprising:
one or more physical processors configured by machine-readable instructions to:
create a training set comprising material network information and network parameter information for a period of time, wherein:
the material network information defines a network for a set of previously synthesized materials and comprises discovery information defining points in time individual previously synthesized materials in the network were first synthesized, and
the network parameter information for the period of time specifies sets of parameter values of the previously synthesized materials; and

use the training set to train a classifier model from the individual previously synthesized materials using the discovery information and network parameter information to generate probabilities for materials being successfully synthesized at individual discrete points in time within the period of time.

2. The system of claim 1 , wherein the discovery information specifies whether individual previously synthesized materials were synthesized at given points in time over the period of time.

3. The system of claim 1 , wherein:
the material network information includes node information characterizing previously synthesized materials represented by nodes in the network; and
the network at discrete points in time include a first point in time, and the network at the first point in time is characterized by the material network information for the nodes in the network representing previously synthesized materials that were synthesized prior to the first point in time.

4. The system of claim 1 , wherein a set of parameter values includes values for at least one of degree parameter, degree centrality parameter, eigenvector centrality parameter, mean shortest path parameter, mean degree of neighbors parameter, and clustering coefficient parameter.

5. The system of claim 1 , wherein the one or more physical processors are further configured by machine-readable instructions to:
update the training set to further comprise synthesis information, wherein the synthesis information defines one or more synthesized materials not included in the network with the previously synthesized materials; and
re-train the classifier model using the updated training set to generate probabilities for materials being successfully synthesized at the individual discrete points in time within the time period.

6. The system of claim 1 , wherein the trained classifier model is stored in a non-transient electronic storage.

7. A method for training a classifier model to determine a probability of successful synthesis of a material at one or more points in time, the method comprising:
creating a training set comprising material network information and network parameter information for a period of time, wherein:
the material network information defines a network for a set of previously synthesized materials and comprises discovery information defining points in time individual previously synthesized materials in the network were first synthesized, and
the network parameter information for the period of time specifies sets of parameter values of the previously synthesized materials; and

using the training set to train a classifier model to generate probabilities for materials being successfully synthesized at individual discrete points in time within the period of time.

8. The method of claim 7 , wherein the discovery information specifies whether individual previously synthesized materials were synthesized at given points in time over the period of time.

9. The method of claim 7 , wherein:
the material network information includes node information characterizing previously synthesized materials represented by nodes in the network; and
the network at discrete points in time include a first point in time, and the network at the first point in time is characterized by the material network information for the nodes in the network representing previously synthesized materials that were synthesized prior to the first point in time.

10. The method of claim 7 , wherein a set of parameter values includes values for at least one of degree parameter, degree centrality parameter, eigenvector centrality parameter, mean shortest path parameter, mean degree of neighbors parameter, and clustering coefficient parameter.

11. The method of claim 7 , further comprising:
updating the training set to further comprise synthesis information, wherein the synthesis information defines one or more synthesized materials not included in the network with the previously synthesized materials; and
re-training the classifier model using the updated training set to generate probabilities for materials being successfully synthesized at the individual discrete points in time within the time period.

12. The method of claim 7 , wherein the trained classifier model is stored in a non-transient electronic storage.

Assignees

• Toyota Res Inst Inc

Inventors

• Aykol Muratahan
• Suram Santosh Karthik
• Hung Linda
• Herring Patrick Kenichi

Classifications

• G06N3/09

  Supervised learning · CPC title

• G06N7/01

  Probabilistic graphical models, e.g. probabilistic networks · CPC title

• G06F18/241

  relating to the classification model, e.g. parametric or non-parametric approaches · CPC title

• G16C20/70

  Machine learning, data mining or chemometrics · CPC title

• G16C20/10

  Analysis or design of chemical reactions, syntheses or processes · CPC title