Temporal supply-related forecasting using artificial intelligence techniques

Patent metadata

Field	Value
Publication number	US-12524736-B2
Application number	US-202218078363-A
Country	US
Kind code	B2
Filing date	Dec 9, 2022
Priority date	Dec 9, 2022
Publication date	Jan 13, 2026
Grant date	Jan 13, 2026

Abstract

Methods, apparatus, and processor-readable storage media for temporal supply-related forecasting using artificial intelligence techniques are provided herein. An example computer-implemented method includes determining one or more forecasts pertaining to supply of at least one item by processing supply-related data using one or more artificial intelligence techniques; generating, based at least in part on the one or more forecasts, one or more temporal recommendations associated with one or more orders of at least one a portion of the at least one item; and performing one or more automated actions based at least in part on the one or more temporal recommendations.

First claim

What is claimed is:

1 . A computer-implemented method comprising:
processing supply-related data associated with at least one item by merging, using one or more artificial intelligence techniques, (i) one or more feature covariates related to the at least one item and (ii) one or more temporal-based covariates into at least one multivariate time series object; determining one or more probabilistic forecasts pertaining to supply of the at least one item by computing, via processing at least a portion of the at least one multivariate time series object using at least one neural network comprising a quantile loss function, a respective prediction value at each of multiple time steps related to the supply of the at least one item; generating, based at least in part on the one or more probabilistic forecasts, one or more temporal recommendations associated with one or more orders of at least a portion of the at least one item; and performing one or more automated actions based at least in part on the one or more temporal recommendations, wherein performing one or more automated actions comprises automatically initiating, by transmitting a set of executable instructions to at least one system associated with one or more automated temporally related asset workflow execution tasks, an increase in a supply of the at least a portion of the at least one item at one or more designated instances of time based at least in part on the one or more temporal recommendations; wherein the method is performed by at least one processing device comprising a processor coupled to a memory.

2 . The computer-implemented method of claim 1 , wherein processing supply-related data comprises processing supply-related data using at least one neural basis expansion analysis for interpretable time series forecasting technique.

3 . The computer-implemented method of claim 2 , wherein processing supply-related data using at least one neural basis expansion analysis for interpretable time series forecasting technique comprises implementing a univariate time series prediction architecture comprising a deep neural network structure based at least in part on forward and backward residual connections and a deep fully connected layer stack.

4 . The computer-implemented method of claim 2 , wherein processing supply-related data using at least one neural basis expansion analysis for interpretable time series forecasting technique comprises configuring one or more hyperparameters of the at least one neural basis expansion analysis for interpretable time series forecasting technique, wherein the one or more hyperparameters comprise at least one of input layer size, output layer size, number of blocks within each of one or more stacks, width of each of one or more fully connected layers in each block of each of one or more stacks, batch size, and epoch information.

5 . The computer-implemented method of claim 2 , wherein processing supply-related data using at least one neural basis expansion analysis for interpretable time series forecasting technique further comprises using one or more quantile regression techniques.

6 . The computer-implemented method of claim 1 , wherein performing one or more automated actions comprises automatically outputting one or more commitment values, based at least in part on the one or more temporal recommendations, to at least one user in response to at least a portion of the one or more orders.

7 . The computer-implemented method of claim 1 , wherein performing one or more automated actions comprises automatically training at least a portion of the one or more artificial intelligence techniques using feedback related to at least a portion of the one or more temporal recommendations.

8 . The computer-implemented method of claim 1 , wherein processing supply-related data using one or more intelligence techniques comprises processing historical values pertaining to supply of the at least one item over one or more temporal periods.

9 . The computer-implemented method of claim 1 , wherein generating one or more temporal recommendations comprises processing, using one or more rule-based techniques, at least a portion of the one or more probabilistic forecasts in conjunction with information pertaining to at least a portion of the one or more orders.

10 . A non-transitory processor-readable storage medium having stored therein program code of one or more software programs, wherein the program code when executed by at least one processing device causes the at least one processing device:
to process supply-related data associated with at least one item by merging, using one or more artificial intelligence techniques, (i) one or more feature covariates related to the at least one item and (ii) one or more temporal-based covariates into at least one multivariate time series object; to determine one or more probabilistic forecasts pertaining to supply of the at least one item by computing, via processing at least a portion of the at least one multivariate time series object using at least one neural network comprising a quantile loss function, a respective prediction value at each of multiple time steps related to the supply of the at least one item; to generate, based at least in part on the one or more probabilistic forecasts, one or more temporal recommendations associated with one or more orders of at least a portion of the at least one item; and to perform one or more automated actions based at least in part on the one or more temporal recommendations, wherein performing one or more automated actions comprises automatically initiating, by transmitting a set of executable instructions to at least one system associated with one or more automated temporally related asset workflow execution tasks, an increase in a supply of the at least a portion of the at least one item at one or more designated instances of time based at least in part on the one or more temporal recommendations.

11 . The non-transitory processor-readable storage medium of claim 10 , wherein processing supply-related data comprises processing supply-related data using at least one neural basis expansion analysis for interpretable time series forecasting technique.

12 . The non-transitory processor-readable storage medium of claim 11 , wherein processing supply-related data using at least one neural basis expansion analysis for interpretable time series forecasting technique comprises implementing a univariate time series prediction architecture comprising a deep neural network structure based at least in part on forward and backward residual connections and a deep fully connected layer stack.

13 . The non-transitory processor-readable storage medium of claim 10 , wherein performing one or more automated actions comprises automatically outputting one or more commitment values, based at least in part on the one or more temporal recommendations, to at least one user in response to at least a portion of the one or more orders.

14 . An apparatus comprising:
at least one processing device comprising a processor coupled to a memory; the at least one processing device being configured:
to process supply-related data associated with at least one item by merging, using one or more artificial intelligence techniques, (i) one or more feature covariates related to the at least one item and (ii) one or more temporal-based covariates into at least one multivariate time series object;
to determine one or more probabilistic forecasts pertaining to supply of the at least one item by computing, via processing at least a portion of the at least one multivariate time series object using at lea

Assignees

• Dell Products Lp

Inventors

• Kumar Dhilip S
• Sahoo Sujit Kumar
• Marappa Devaraj
• Pularru Nikhil
• Kotikanyadanam Aravinda Rajagopal
• Kalyani Rajeshwari

Classifications

• G06Q30/0202 Primary

  Market predictions or forecasting for commercial activities · CPC title

• G06Q10/087 Primary

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