Multi-Agent Generative Adversarial Imitative Superlearning

1 . A method for improving simulation-based question answering, the method comprising:

using a resource intensive algorithm to answer a first question of a question type;

generating at least one training example from the normal operation of the resource intensive algorithm;

training a lightweight machine learning model based on the at least one training example to produce answers to questions of the question type; and

using the lightweight machine learning model to produce an answer to a second question of the question type.

2 . The method of claim 1 , wherein the resource intensive algorithm answers the first question by tuning at least one or more input of a simulator until an output of the simulator sufficiently meets a criteria of the question.

3 . The method of claim 2 , wherein the at least one training example includes:

output of the simulation as input, and

input to the simulation as output.

4 . The method of claim 1 , wherein the resource intensive algorithm is a multi-agent optimizer.

5 . The method of claim 4 , further comprising:

using the answer to the second question as a starting position of at least one agent of the multi-agent optimizer; and

using the multi-agent optimizer to produce a refined answer to the second question.

6 . The method of claim 4 , wherein generating at least one training example comprises generating a training example from the location of an agent after each optimization iteration of the multi-agent optimizer.

7 . The method of claim 1 , further comprising:

using the answer to the second question as input to a simulator to produce a simulated result, and

using the simulated result to verify acceptability of the answer to the second question.

8 . The method of claim 1 , further comprising:

using the answer to the second question to perform at least one control action in a real world system.

9 . The method of claim 1 , wherein the resource intensive algorithm utilizes a digital twin of a real world system to answer the first question.

10 . The method of claim 9 , wherein using a resource intensive algorithm to answer a first question of a question type comprises computing a cost function from the digital twin.

11 . The method of claim 1 , further comprising:

using the resource intensive algorithm to answer a third question of a different question type;

generating at least one additional training example from the normal operation of the resource intensive algorithm;

training an additional lightweight machine learning model based on the at least one additional training example to produce answers to questions of the different question type; and

using the additional lightweight machine learning model to produce an answer to a fourth question of the different question type.

12 . A controller that utilizes a simulation in controlling a real world system, the controller comprising a memory and a processer, wherein the processor is in communication with the memory configured to:

use a resource intensive algorithm to answer a first question of a question type;

generate at least one training example from the normal operation of the resource intensive algorithm;

train a lightweight machine learning model based on the at least one training example to produce answers to questions of the question type; and

use the lightweight machine learning model to produce an answer to a second question of the question type.

13 . The controller of claim 12 , further comprising the processor configured to:

use the resource intensive algorithm to answer a third question of a different question type;

generate at least one additional training example from the normal operation of the resource intensive algorithm;

train an additional lightweight machine learning model based on the at least one additional training example to produce answers to questions of the different question type; and

use the additional lightweight machine learning model to produce an answer to a fourth question of the different question type.

14 . The controller of claim 12 , wherein the resource intensive algorithm answers the first question by tuning at least one or more input of a simulator until an output of the simulator sufficiently meets a criteria of the question.

15 . The controller of claim 14 , wherein the at least one training example includes:

output of the simulation as input, and

input to the simulation as output.

16 . The controller of claim 12 , wherein the resource intensive algorithm is a multi-agent optimizer.

17 . The controller of claim 16 , further comprising the processor configured to:

use the answer to the second question as a starting position of at least one agent of the multi-agent optimizer; and

use the multi-agent optimizer to produce a refined answer to the second question.

18 . The controller of claim 16 , wherein generating at least one training example comprises generating a training example from the location of an agent after each optimization iteration of the multi-agent optimizer.

19 . A non-transitory machine-readable medium encoded with instructions for execution by a processor, the non-transitory machine-readable medium comprising:

instructions for using a resource intensive algorithm to answer a first question of a question type;

instructions for generating at least one training example from the normal operation of the resource intensive algorithm;

instructions for training a lightweight machine learning model based on the at least one training example to produce answers to questions of the question type; and

instructions for using the lightweight machine learning model to produce an answer to a second question of the question type.

20 . The non-transitory machine-readable medium of claim 19 wherein the resource intensive algorithm utilizes a digital twin of a real world system to answer the first question.