Risk adaptive asset management

Patent metadata

Field	Value
Publication number	US-12248989-B2
Application number	US-202318541771-A
Country	US
Kind code	B2
Filing date	Dec 15, 2023
Priority date	Dec 9, 2021
Publication date	Mar 11, 2025
Grant date	Mar 11, 2025

Abstract

A computer-implemented method is provided for determining an action with respect to a given portfolio of items for supply chain management. The method includes acquiring, by a hardware processor, a feature vector for supply chain delivery trends, the given portfolio, and a current investment amount. The method further includes determining, by the hardware processor, whether a current supply chain delivery situation is normal or abnormal based on the feature vector. The method also includes performing a risk-avoidance action to reduce the current investment amount and avoid potential supply chain delivery losses, responsive to a determination that the current supply chain delivery situation is abnormal. The method additionally includes performing a risk adaptive action to increase the current investment amount and incur potential supply chain delivery gains by using a distributional reinforcement learning process, responsive to a determination that the current supply chain delivery situation is normal.

First claim

The invention claimed is:

1. A computer-implemented method for determining an action with respect to a given portfolio of items for supply chain management, comprising:
determining, by a hardware processor, whether a current supply chain delivery situation is normal or abnormal based on a feature vector for supply chain delivery trends and a current investment amount;
performing a risk-avoidance action to reduce the current investment amount, responsive to a determination that the current supply chain delivery situation is abnormal; and
performing a risk adaptive action to increase the current investment amount, responsive to a determination that the current supply chain delivery situation is normal, including dispatching and controlling additional autonomous delivery vehicles.

2. The computer-implemented method of claim 1 , wherein performing a risk adaptive action includes:
estimating a return distribution for each of candidate actions by using a distributional reinforcement learning process with a Laplace distribution to obtain an estimated return distribution;
selecting and performing an action that optimizes a risk-avoidance criterion of the estimated return distribution, responsive to the current investment amount exceeding a threshold; and
selecting and performing an action that optimizes a risk-seeking criterion of the estimated return distribution, responsive to the current investment amount not exceeding the threshold.

3. The computer-implemented method of claim 2 , wherein estimating a return distribution for each of candidate actions by using the distributional reinforcement learning process with a Laplace distribution comprises learning parameters of a Laplace distribution model with return samples.

4. The computer-implemented method of claim 2 , further comprising modeling a tail of the supply chain delivery trends using the Laplace distribution.

5. The computer-implemented method of claim 2 , wherein performing an action that optimizes a risk-avoidance criterion of the estimated return distribution comprises making an investment corresponding to a fraction of the current investment amount corresponding to a given maximum acceptable loss amount.

6. The computer-implemented method of claim 2 , wherein performing an action that optimizes risk-seeking criterion of the estimated return distribution comprises making an investment corresponding to a full amount of the current investment amount.

7. The computer-implemented method of claim 1 , wherein performing a risk-avoidance action comprises optimizing a Sharpe ratio of the current investment amount.

8. The computer-implemented method of claim 1 , wherein performing a risk-avoidance action comprises selecting an action that optimizes an average return−(λ×a standard deviation), where λ is a tunable parameter greater than 0.

9. The computer-implemented method of claim 1 , further comprising avoiding likely delivery delays by intentionally delaying a scheduled purchase of a commodity amount until the current supply chain delivery situation changes from abnormal to normal to minimize a time in transit.

10. The computer-implemented method of claim 1 , further comprising maintaining perishable-based inventory control to automatically discard, by a garbage disposal mechanism, products past an expiration date.

11. The computer-implemented method of claim 1 , wherein performing a risk adaptive action to increase the current investment amount includes purchasing and introducing an amount of product into commerce to exploit the determination that the current supply chain delivery situation is normal.

12. The computer-implemented method of claim 1 , wherein performing a risk adaptive action to increase the current investment amount includes dispatching and controlling additional autonomous vehicles to handle an increase in supply chain deliveries.

13. The computer-implemented method of claim 1 , wherein the feature vector comprises a ratio of missed deliveries versus scheduled deliveries.

14. The computer-implemented method of claim 1 , wherein performing an action that optimizes a risk-seeking criterion comprises dispatching and controlling a fleet of autonomous vehicles in an inclement weather situation.

15. A computer program product for determining an action with respect to a given portfolio of items for supply chain management, the computer program product comprising a non-transitory computer readable storage medium having program instructions embodied therewith, the program instructions executable by a computer to cause the computer to perform a computer-implemented method comprising:
determining, by a hardware processor, whether a current supply chain delivery situation is normal or abnormal based on a feature vector for supply chain delivery trends and a current investment amount;
performing a risk-avoidance action to reduce the current investment amount, responsive to a determination that the current supply chain delivery situation is abnormal; and
performing a risk adaptive action to increase the current investment amount using a distributional reinforcement learning process, responsive to a determination that the current supply chain delivery situation is normal, including dispatching and controlling additional autonomous delivery vehicles.

16. The computer-implemented method of claim 15 , wherein performing a risk adaptive action includes:
estimating a return distribution for each of candidate actions by using the distributional reinforcement learning process with a Laplace distribution to obtain an estimated return distribution;
performing an action that optimizes a risk-avoidance criterion of the estimated return distribution, responsive to the current investment amount exceeding a threshold; and
performing an action that optimizes a risk-seeking criterion of the estimated return distribution, responsive to the current investment amount not exceeding the threshold.

17. The computer-implemented method of claim 16 , wherein estimating a return distribution for each of candidate actions by using the distributional reinforcement learning process with a Laplace distribution comprises learning parameters of a Laplace distribution model with return samples.

18. The computer-implemented method of claim 16 , further comprising modeling a tail of the supply chain delivery trends using the Laplace distribution.

19. The computer-implemented method of claim 16 , wherein performing an action that optimizes a risk-avoidance criterion of the estimated return distribution comprises making an investment corresponding to a fraction of the current investment amount corresponding to a given maximum acceptable loss amount.

20. The computer-implemented method of claim 16 , wherein performing an action that optimizes risk-seeking criterion of the estimated return distribution comprises making an investment corresponding to a full amount of the current investment amount.

21. The computer-implemented method of claim 15 , wherein performing a risk-avoidance action comprises optimizing a Sharpe ratio of the current investment amount.

22. The computer-implemented method of claim 15 , wherein performing an action that optimizes a risk-seeking criterion comprises dispatching and controlling a fleet of autonomous vehicles in an inclement weather situation.

23. The computer-implemented method of claim 15 , further comprising avoiding likely delivery delays by intentionally delaying a scheduled purchase of a commodity amount u

Assignees

• IBM

Inventors

• Morimura Tetsuro

Classifications

• G06N20/00

  Machine learning · CPC title

• G06Q10/0833 Primary

  Tracking · CPC title

• G06Q10/06315

  Needs-based resource requirements planning or analysis · CPC title

• G06Q10/087

  Inventory or stock management, e.g. order filling, procurement or balancing against orders · CPC title

• G06Q40/06 Primary

  Asset management; Financial planning or analysis · CPC title