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-2019378029-A1
Application number	US-201816004232-A
Country	US
Kind code	A1
Filing date	Jun 8, 2018
Priority date	Jun 8, 2018
Publication date	Dec 12, 2019
Grant date	—

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:

1 . A system configured to determine a probability of successful synthesis of a material in the real world at one or more points in time, the system comprising:
one or more physical processors configured by machine-readable instructions to:
obtain material network information defining a network for a set of previously synthesized materials, the material network information including node information, edge information, and discovery information, the node information characterizing previously synthesized materials represented by nodes in the network, the edge information representing connections between the previously synthesized materials based on shared components between individual ones of the previously synthesized materials, wherein the edge information further represents connections between the previously synthesized materials based on other relationship between the previously synthesized materials including relationships derived from thermodynamic co-existence information and/or phase diagrams, the discovery information defining points in time the individual previously synthesized materials in the network were first synthesized in the real world;
determine network parameter information for the network of the period of time, the network parameter information specifying sets of parameter values of the previously synthesized materials, wherein the sets of parameter values of the previously synthesized materials are determined by representing the previously synthesized materials as individual nodes in the network at discrete points in time over a period of time and determined the sets of parameter values based on the network, wherein the network at a given discrete points in time over the period of time includes the nodes representing the previously synthesized materials that were synthesized by the given discrete points in time, and the previously synthesized materials that were not synthesized by the given discrete points in time are independently introduced in the network to determine the sets of parameter values of the previously synthesized materials that were not synthesized;
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 in the real world at the individual discrete points in time;
determine a probability of an unsynthesized hypothetical material being successfully synthesized in the real world at the individual discrete points in time by applying the classifier model to unsynthesized material information defining the unsynthesized hypothetical material.

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

3 . The system of claim 1 , wherein 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 representing previously synthesized materials that were synthesized in the real world prior to the first point in time.

4 . The system of claim 1 , wherein a set of parameter values include values for one or more of degree parameter, degree centrality parameter, eigenvector centrality parameter, mean shortest path parameter, mean degree of neighbors parameter, and/or clustering coefficient parameter.

5 . The system of claim 1 , wherein the one or more physical processors are further configured by machine-readable instructions to:
obtain synthesis information, the synthesis information defining one or more synthesized materials not included in the network with the previously synthesized materials; modify the material network information to include the synthesized materials in the network, the modification of the material network information include modifying the node information to represent the synthesized materials as additional nodes to the network, modifying the edge information to represent connections between the synthesized materials and previously synthesized materials, modifying the discovery information to include points in time the synthesized materials were first synthesized in the real world; determine the network parameter information for the network over a period of time, the network parameter information specifying sets of parameter values of the previously synthesized materials and the synthesized materials, wherein the sets of parameter values of the previously synthesized materials and the synthesized materials are determined by representing the previously synthesized materials and the synthesized materials as individual nodes in the network at discrete points in time over a period of time and determined the sets of parameter values based on the network, wherein the network at a given discrete points in time over the period of time includes the nodes representing the previously synthesized materials and the synthesized materials that were synthesized by the given discrete points in time, and the previously synthesized materials and the synthesized materials that were not synthesized by the given discrete points in time are independently introduced in the network to determine the sets of parameter values of the previously synthesized materials and the synthesized materials that were not synthesized; train the classifier model using the discovery information and network parameter information to generate probabilities for materials being successfully synthesized in the real world at the individual discrete points in time; and determine a probability of the unsynthesized hypothetical material being successfully synthesized in the real world at the individual discrete points in time by applying the classifier model to unsynthesized material information defining the unsynthesized hypothetical material.

6 . The system of claim 1 , wherein the one or more physical processors are further configured by machine-readable instructions to:
determine a change in the probability that the unsynthesized hypothetical material will be successfully synthesized in the real world at the individual discrete points in time based on synthesis information; and provide a user with the change in the probability that the unsynthesized hypothetical material will be successfully synthesized in the real world at the individual discrete points in time on an interface.

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

8 . A system configured to determine a probability of successful synthesis of a material in the real world at one or more points in time, the system comprising:
a non-transient electronic storage configured to store a trained classifier model; one or more physical processors configured by machine-readable instructions to:
obtain unsynthesized hypothetical material information, the unsynthesized hypothetical material information defining a set of parameter values for an unsynthesized hypothetical material;
obtaining the classifier model from the non-transient electronic storage, the classifier model having been trained by a training data set, the training data set include (i) network parameter information and (ii) discovery information;
determine probabilities of successful synthesis of the unsynthesized hypothetical material at discrete points in time of the period of time by applying the classifier model to the unsynthesized hypothetical material information;
provide a user with the probabilities of successful synthesis of the unsynthesized hypothetical material at discrete points in time of the period of time on an interface

Assignees

• Toyota Res Inst Inc

Inventors

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

Classifications

• G16C60/00 Primary

  Computational materials science, i.e. ICT specially adapted for investigating the physical or chemical properties of materials or phenomena associated with their design, synthesis, processing, characterisation or utilisation · CPC title

• G06N7/01

  Probabilistic graphical models, e.g. probabilistic networks · 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

• G06F18/241

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