Determining causal models for controlling environments

Patent metadata

Field	Value
Publication number	US-12422797-B2
Application number	US-202418769969-A
Country	US
Kind code	B2
Filing date	Jul 11, 2024
Priority date	Mar 15, 2019
Publication date	Sep 23, 2025
Grant date	Sep 23, 2025

Abstract

Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for determining causal models for controlling environments. One of the methods includes identifying a procedural instance; selecting control settings for the procedural instance, comprising, for a particular one of the controllable elements: assigning the procedural instance to a cluster for the particular controllable element in accordance with current values of a set of clustering parameters for the particular controllable element; and selecting a setting for the particular controllable element for the procedural instances based on a causal model that is specific to the cluster; obtaining environment responses to the selected control settings that define a value of the performance metric for the procedural instance; and updating, for the particular controllable element, the causal model for the cluster for the controllable element to which the procedural instance was assigned based on the value of the performance metric.

First claim

What is claimed is:

1. A method of controlling an environment by selecting control settings that include a respective setting for each of a plurality of controllable elements of the environment, the method comprising repeatedly performing the following:
identifying a procedural instance, wherein the procedural instance is a segment of the environment to which control settings are to be applied;
selecting control settings for the procedural instance, comprising, for a particular one of the controllable elements:
assigning the procedural instance to a cluster for the particular controllable element in accordance with current values of a set of clustering parameters for the particular controllable element;
determining that optimal settings for one of the controllable elements or for one of the internal parameters are approaching a boundary of the search space; and
receiving authorization input from a user, based upon a request displayed to the user, to expand the space of possible values for the controllable element for the cluster to which the procedural instance was assigned.

2. The method of claim 1 , wherein the clustering parameters define one or more hyperparameters of a clustering technique used to assign the procedural instance to a cluster.

3. The method of claim 2 , wherein the particular controllable element is associated with a data inclusion window that defines how many previous procedural instances are considered by the clustering technique, and wherein the method further comprises:
determining, based on the current values of the set of clustering parameters, current values for each of the one or more hyperparameters;
clustering, based on the current values of each of the one or more hyperparameters, the previous procedural instances that are within the data inclusion window into a plurality of clusters, and
generating, for each of the plurality of clusters, a causal model that is specific to the cluster from the previous procedural instances that are clustered into the cluster and the environment responses for the previous procedural instances.

4. The method of claim 3 , wherein the clustering technique clusters the previous environment responses based on environment characteristics of the procedural instances.

5. The method of claim 4 , further comprising:
stochastically varying the environmental characteristics that are considered by the clustering technique.

6. The method of claim 3 , wherein clustering the previous procedural instances comprises:
determining, based on the current values of the clustering parameters, a constraint on a cluster size for the clusters; and
clustering the previous procedural instances into the plurality of clusters such that the clusters have maximum within-cluster similarity of the impact of controllable elements on the performance metric and maximum between-cluster difference in the impact of controllable elements on the performance metric subject to the constraint on cluster size.

7. The method of claim 6 , wherein determining the constraint on cluster size comprises:
sampling a cluster size from a range of possible cluster sizes defined by the current values of the clustering parameters.

8. The method of claim 7 , further comprising:
identifying, based on a statistical power analysis, an adjusted range of possible cluster sizes that result in a desired range of precision of the causal models for the clusters; and
adjusting the current values of the clustering parameters to define the adjusted range.

9. The method of claim 7 , further comprising:
adjusting, based on a measure of system performance, a frequency with which each possible cluster size is sampled from the range.

10. The method of claim 2 , further comprising:
adjusting the data inclusion window for the particular controllable element.

11. A system comprising one or more computers and one or more storage devices storing instructions that when executed by the one or more computers cause the one or more computers to repeatedly perform the following:
identify a procedural instance, wherein the procedural instance is a segment of the environment to which control settings are to be applied;
select control settings for the procedural instance, comprising, for a particular one of the controllable elements:
assign the procedural instance to a cluster for the particular controllable element in accordance with current values of a set of clustering parameters for the particular controllable element;
determine that optimal settings for one of the controllable elements or for one of the internal parameters are approaching a boundary of the search space; and
receive authorization input from a user, based upon a request displayed to the user, to expand the space of possible values for the controllable element for the cluster to which the procedural instance was assigned.

12. One or more non-transitory computer-readable storage media storing instructions that when executed by one or more computers cause the one or more computers to repeatedly perform the following:
identifying a procedural instance, wherein the procedural instance is a segment of the environment to which control settings are to be applied;
selecting control settings for the procedural instance, comprising, for a particular one of the controllable elements:
assigning the procedural instance to a cluster for the particular controllable element in accordance with current values of a set of clustering parameters for the particular controllable element;
determining that optimal settings for one of the controllable elements or for one of the internal parameters are approaching a boundary of the search space; and
receiving authorization input from a user, based upon a request displayed to the user, to expand the space of possible values for the controllable element for the cluster to which the procedural instance was assigned.

13. The one or more non-transitory computer-readable storage media of claim 12 , wherein the clustering parameters define one or more hyperparameters of a clustering technique used to assign the procedural instance to a cluster.

14. The one or more non-transitory computer-readable storage media of claim 13 , wherein the particular controllable element is associated with a data inclusion window that defines how many previous procedural instances are considered by the clustering technique, and wherein the method further comprises:
determining, based on the current values of the set of clustering parameters, current values for each of the one or more hyperparameters;
clustering, based on the current values of each of the one or more hyperparameters, the previous procedural instances that are within the data inclusion window into a plurality of clusters, and
generating, for each of the plurality of clusters, a causal model that is specific to the cluster from the previous procedural instances that are clustered into the cluster and the environment responses for the previous procedural instances.

15. The one or more non-transitory computer-readable storage media of claim 14 , wherein the clustering technique clusters the previous environment responses based on environment characteristics of the procedural instances.

16. The one or more non-transitory computer-readable storage media of claim 14 , wherein clustering the previous procedural instances comprises:
determining, based on the current values of the clustering parameters, a constraint on a cluster size for the clusters; and
clustering the previous procedural instances into the plurality of clusters such that the clusters have maximu

Assignees

• 3M Innovative Properties Company

Inventors

• Nayar Himanshu
• Arsenault Frederick J
• Johnson Nicholas A
• Brooks Brian E
• Benoit Gilles J
• Olson Peter O
• Olson Tyler W

Classifications

• G05B23/0248

  Causal models, e.g. fault tree; digraphs; qualitative physics · CPC title

• G06N7/01

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

• G06F18/2193

  based on specific statistical tests · CPC title

• G05B13/021

  in which a variable is automatically adjusted to optimise the performance · CPC title

• B60W40/105

  Speed · CPC title