US 12,407,627 B2
Systems and methods for conserving computational resources by reducing network traffic predictions via machine learning models
Ateet Kumar Awasthi, McKinney, TX (US); Chris Fields, Roanoke, TX (US); Saral Jain, McKinney, TX (US); Matt Howarth, Garland, TX (US); and Vedhasree Periathambi, Irving, TX (US)
Assigned to Capital One Services, LLC, McLean, VA (US)
Filed by Capital One Services, LLC, McLean, VA (US)
Filed on Jan. 26, 2024, as Appl. No. 18/424,509.
Prior Publication US 2025/0247339 A1, Jul. 31, 2025
Int. Cl. G06F 15/173 (2006.01); H04L 43/106 (2022.01); H04L 43/16 (2022.01); H04L 47/80 (2022.01); H04L 47/83 (2022.01)
CPC H04L 47/83 (2022.05) [H04L 43/106 (2013.01); H04L 43/16 (2013.01); H04L 47/801 (2013.01)] 20 Claims
OG exemplary drawing
 
1. A system for conserving computational resources by reducing network traffic predictions via machine learning models facilitating cloud computing system resource allocations, the system comprising:
one or more processors executing computer program instructions that, when executed, cause operations comprising:
in connection with automated periodic triggering of network traffic predictions via a first machine learning model at scheduled intervals, receiving, via the first machine learning model, an initial network traffic prediction for a network service executing within a cloud computing environment, and performing a first task allocation process for the network service based on the initial network traffic prediction, the first task allocation process allocating a set of instantiated tasks to the network service, the set of instantiated tasks being instantiated within the cloud computing environment;
prior to a next scheduled interval of the automated periodic triggering, detecting whether an amount of actual network traffic experienced by the network service (i) exceeds an upper bound of the initial network traffic prediction and (ii) continuously satisfies a relative traffic threshold for a threshold time period, the relative traffic threshold being a different parameter from the upper bound and is relative the initial network traffic prediction; and
in response to the amount of actual network traffic exceeding the upper bound of the initial network traffic prediction and continuously satisfying the relative traffic threshold for the threshold time period, executing a machine learning model to generate a subsequent network traffic prediction for the network service, and performing a second task allocation process for the network service based on the subsequent network traffic prediction, the second task allocation process allocating an additional set of instantiated tasks to the network service,
wherein the machine learning model is not executed, to generate the subsequent network traffic prediction, in response to the amount of actual network traffic failing to exceed the upper bound of the initial network traffic prediction and failing to continuously satisfy the relative traffic threshold for the threshold time period.