Granted Patent B2

US 12,614,197 · App. 17/401,538 · Granted Apr 28, 2026

Inventors: Mary Xiaoyu Ma (Oakville, CA); Joel Aidan Gritter (Vineland Station, CA); Inaara Hasmani (Toronto, CA); Nicholas Andrien Prayogo (Toronto, CA); Imran Habib (Toronto, CA); Richard Stanton (Toronto, CA); Jenna Hague (Mississauga, CA); Victor Cheng (Toronto, CA)

Assignee: ROYAL BANK OF CANADA

G06Q30/0201G06F18/217G06N20/00G06Q30/0202G06Q30/0207G06Q20/389

Abstract

Claims (62)

1 . A system for intelligent resource management, the system comprising:

at least one processor; and

a memory comprising instructions which, when executed by the processor, configure the processor to:

receive resource data and usage data, wherein the usage data comprises location data corresponding to the resource data;

preprocess and convert the resource data and the usage data into operational data;

train a foresight model on labelled data, wherein the foresight model comprises an autoregressive model that iteratively generates forecasts at time intervals in a time series, the autoregressive model comprising a set of neurons defining a plurality of hidden states, wherein the plurality of hidden states encapsulates historical data of a plurality of users, wherein the autoregressive model updates the set of neurons based on whether each forecast matches a truth future data;

update the foresight model using the operational data;

receive a forecast generated by the foresight model for the time series, the forecast associated with a future resource state or event associated with the operational data for a location identified in the location data;

send a notification for a recommended action based on the forecast, wherein the recommended action optimizes the future resource state or event associated with the operational data to reach a target outcome state or event;

cause to display, at an electronic device, a graphical interface comprising a first graphical user interface (GUI) element displaying a location view of the generated forecast for the location identified in the location data and a second GUI element comprising an input control, wherein data from the input control of the second GUI element is used to update the displayed forecast in the first GUI element;

receive, from the electronic device, user input representing a modification of the operational data, the modification of the operational data comprising a modification to the time series, the user input received from the input control of the second GUI element;

receive an updated forecast generated by the foresight model, the forecast associated with a future resource state or event associated with the modified operational data for the same location identified in the location data, wherein the updated forecast is based on the user input received from the second GUI element; and

in response to the user input and based on the modification of the operational data, cause to display, at the electronic device, the updated forecast in the first GUI element, wherein the first GUI element displays the updated forecast in a location view corresponding to the same location identified in the location data.

2 . The system as claimed in claim 1 , wherein the at least one processor is configured to initially train the model.

3 . The system as claimed in claim 1 , wherein the recommended action is based on the forecast being below a threshold.

4 . The system as claimed in claim 1 , wherein the recommended action is based on a match between the recommended action and the forecast.

5 . The system as claimed in claim 1 , wherein the at least one processor is configured to send the notification based on a current time and a location of a market associated with the resource data.

6 . The system as claimed in claim 1 , wherein:

the resource data comprises transactional data;

the usage data comprises at least one of sales data, customer data, location data, competitor data, and inventory data; and

the recommended action comprises an offer or sales promotion.

7 . The system as claimed in claim 1 , wherein:

the usage data comprises at least one of transactional data, merchant information, or client card information; and

the resource data comprises at least one of a commercial account number, or a client identifier.

8 . The system as claimed in claim 7 , wherein to preprocess the resource data and the usage data into operational data, the at least one process of configured to:

merge the resource data and the usage data; and

enhance the merged data with external context data.

9 . The system as claimed in claim 8 , wherein the at least one processor is configured to at least one of:

maintain the commercial account number intact after the merge of the resource data and the usage data;

maintain a purchase amount of the transactional data intact after the merge of the resource data and the usage data; or

link the transactional data to the commercial account number.

10 . The system as claimed in claim 1 , wherein the at least one processor is configured to train a recommendation model to provide the recommendation action based on the operational data using at least one of stochastic optimization or reinforcement learning.

11 . A computer-implemented method for intelligent resource management, the method comprising:

receiving resource data and usage data, wherein the usage data comprises location data corresponding to the resource data;

preprocessing and converting the resource data and the usage data into operational data;

training a foresight model on labelled data, wherein the foresight model comprises an autoregressive model that iteratively generates forecasts at time intervals in a time series, the autoregressive model comprising a set of neurons defining a plurality of hidden states, wherein the plurality of hidden states encapsulates historical data of a plurality of users, wherein the autoregressive model updates the set of neurons based on whether each forecast matches a truth future data;

updating the foresight model using the operational data;

receiving a forecast generated by the foresight model for the time series, the forecast associated with a future resource state or event associated with the operational data for a location identified in the location data;

sending a notification for a recommended action based on the forecast, wherein the recommended action optimizes the future resource state or event associated with the operational data to reach a target outcome state or event;

causing to display, at an electronic device, a graphical interface comprising a first graphical user interface (GUI) element displaying a location view of the generated forecast for the location identified in the location data and a second GUI element comprising an input control, wherein data from the input control of the second GUI element is used to update the displayed forecast in the first GUI element;

receiving, from the electronic device, user input representing a modification of the operational data, the modification of the operational data comprising a modification to the time series, the user input received from the input control of the second GUI element;

receiving an updated forecast generated by the foresight model, the forecast associated with a future resource state or event associated with the modified operational data for the same location identified in the location data, wherein the updated forecast is based on the user input received from the second GUI element; and

in response to the user input and based on the modification of the operational data, causing to display, at the electronic device, the updated forecast in the first GUI element, wherein the first GUI element displays the updated forecast in a location view corresponding to the same location identified in the location data.

12 . The computer-implemented method as claimed in claim 11 , further comprising initially training the model.

13 . The computer-implemented method as claimed in claim 11 , wherein the recommended action is based on the forecast being below a threshold.

14 . The computer-implemented method as claimed in claim 11 , wherein the recommended action is based on a match between the recommended action and the forecast.

15 . The computer-implemented method as claimed in claim 11 , further comprising sending the notification based on a current time and a location of a market associated with the resource data.

16 . The computer-implemented method as claimed in claim 11 , wherein:

the resource data comprises transactional data;

the usage data comprises at least one of sales data, customer data, location data, competitor data, and inventory data; and

the recommended action comprises an offer or sales promotion.

17 . The computer-implemented method as claimed in claim 11 , wherein:

the usage data comprises at least one of transactional data, merchant information, or client card information; and

the resource data comprises at least one of a commercial account number, or a client identifier.

18 . The computer-implemented method as claimed in claim 17 , wherein preprocessing the resource data and the usage data into operational data comprises:

merging the resource data and the usage data; and

enhancing the merged data with external context data.

19 . The computer-implemented method as claimed in claim 18 , comprising at least one of:

maintaining the commercial account number intact after the merge of the resource data and the usage data;

maintaining a purchase amount of the transactional data intact after the merge of the resource data and the usage data; or

linking the transactional data to the commercial account number.

20 . The computer-implemented method as claimed in claim 11 , comprising training a recommendation model to provide the recommendation action based on the operational data using at least one of stochastic optimization or reinforcement learning.

Assignments (1)

Continuity (2)

References Cited (9)