Method and apparatus for reinforcement learning training sessions with consideration of resource costing and resource utilization
Reinforcement learning enables a framework of information technology assets that include software elements, computational hardware assets, and/or, bundled software and computational hardware systems and products. The performance of successive sessions of an inner loop reinforcement learning is directed and monitored by an outer loop reinforcement learning wherein the outer loop reinforcement learning is designed to reduce financial costs and computational asset requirements and/or optimize learning time in successive instantiations of inner loop reinforcement learning training sessions. The framework enables consideration of the license costs of domain specific simulators, the usage cost of hardware platforms, and the progress of a particular reinforcement learning training. The framework further enables reductions of these costs to orchestrate and train a neural network under budget constraints with respect to the available hardware and software licenses available at runtime. These improvements and optimizations may be performed by using heuristics and neural network algorithms.
1. A computer implemented nesting reinforcement machine learning method applied to improve a cost effectiveness of a nested software-encoded reinforcement learning program, the method comprising:
(a.) a nesting reinforcement machine learning software-encoded program performing a plurality of cycles of execution (“executions”) of the nested software-encoded reinforcement learning program;
(b.) the nesting reinforcement machine learning software-encoded program evaluating variations of at least one financial cost value monitored during the executions of the nested software-encoded reinforcement learning program; and
(c.) the nesting reinforcement machine learning software-encoded program adjusting at least one resource parameter value of the nested software-encoded reinforcement learning program at least partly in consideration of the variations of the at least one financial cost value observed during the executions of the nested software-encoded reinforcement learning program.
2. The method of claim 1 , wherein the at least one financial cost value is derived from a license fee.
3. The method of claim 2 , wherein the license fee is incurred in the use of a software program.
4. The method of claim 3 , wherein the software program is a simulation program.
5. The method of claim 4 , wherein the software program is at least partially executed on a prespecified type of computational system.
6. The method of claim 4 , wherein the nested software-encoded reinforcement learning program bi-directionally communicates with a software-encoded model in performing at least one cycle of execution of the nested software-encoded reinforcement learning program.
7. The method of claim 1 , wherein the nested software-encoded reinforcement learning program bi-directionally communicates with a software-encoded model in performing at least one cycle of execution of the nested software-encoded reinforcement learning program.
8. The method of claim 1 , wherein the at least one financial cost value is related to a usage fee of a bundled software and computational hardware system.
9. The method of claim 1 , wherein the nesting reinforcement machine learning software-encoded program adapts to one or more granularity settings available in a domain-specific simulator.
10. The method of claim 1 , wherein the nesting reinforcement machine learning software-encoded program adapts to the licensing model associated with the granularity settings of the simulator.
11. The method of claim 1 , wherein the nesting reinforcement machine learning software-encoded program includes a time limitation of duration of the plurality of cycles of execution of the software-encoded reinforcement learning program.
12. The method of claim 1 , wherein the at least one resource parameter value limits a financial expenditure.
13. The method of claim 1 , wherein the at least one resource parameter value limits a total count of licenses to be accessed in a performance of a succeeding plurality of cycles of execution of the nested software-encoded reinforcement learning program.
14. The method of claim 13 wherein the limitation of the total count of licenses to be accessed in a performance of a succeeding plurality of cycles of execution of the nested software-encoded reinforcement learning program includes at least one software program license.
15. The method of claim 13 , wherein the limitation of the total count of licenses to be accessed in a performance of a succeeding plurality of cycles of execution of the nested software-encoded reinforcement learning program includes at least one software license.
16. The method of claim 15 , wherein the software license permits access to a simulation program.
17. The method of claim 16 , wherein the simulation program is at least partially executed on a computational system accessed via an electronic communications network.
18. The method of claim 13 , wherein the limitation of the total count of licenses to be accessed in a performance a succeeding plurality of cycles of execution of the nested software-encoded reinforcement learning program includes at least one computational system license.
19. A computing system, comprising:
one or more processors; and
a memory coupled to the one or more processors and configured to implement a shared storage system, wherein the memory is further configured to store program instructions executable by the one or more processors to implement:
(a.) a nesting software-encoded reinforcement machine learning program performing a plurality of cycles of execution of a nested software-encoded reinforcement learning program, whereby the nested software-encoded reinforcement learning program is nested within the software-encoded reinforcement machine learning program;
(b.) evaluating at least one financial cost value related to the execution of the nested software-encoded reinforcement learning program; and
(c.) adjusting at least one resource parameter value of the nested software-encoded reinforcement learning program at least partly in consideration of variations of the at least one financial cost value as monitored by the nesting software-encoded reinforcement machine learning program during the plurality of cycles of execution of the nested software-encoded reinforcement learning program.
20. The system of claim 19 , wherein the at least one financial cost value is derived from a license fee.
21. The system of claim 20 , wherein the license fee is incurred in the use of a software program.
22. The system of claim 21 , wherein the software program is a simulation program.