US 12,271,939 B2
Automation engine using a hyperparameter learning model to optimize sub-systems of an online system
Sonali Deepak Chhabria, San Carlos, CA (US); Xiangyu Wang, San Jose, CA (US); Aman Jain, Toronto (CA); Ganesh Krishnan, San Francisco, CA (US); Trace Levinson, San Francisco, CA (US); and Jian Wang, Saratoga, CA (US)
Assigned to Maplebear Inc., San Francisco, CA (US)
Filed by Maplebear Inc., San Francisco, CA (US)
Filed on Jun. 30, 2022, as Appl. No. 17/855,788.
Prior Publication US 2024/0005381 A1, Jan. 4, 2024
Int. Cl. G06Q 30/0601 (2023.01); G01C 21/34 (2006.01); G06Q 10/0631 (2023.01); G06Q 10/0637 (2023.01); G06Q 10/087 (2023.01)
CPC G06Q 30/0635 (2013.01) [G01C 21/3407 (2013.01); G06Q 10/06311 (2013.01); G06Q 10/06375 (2013.01); G06Q 10/087 (2013.01)] 17 Claims
OG exemplary drawing
 
1. A method comprising:
obtaining marketplace state data associated with an online concierge system that facilitates processing of a request, from a customer via a customer application, for procurement of one or more items from one or more warehouses, assignment of the request to an available shopper via a shopper application, and generation of routing instructions for delivery of the one or more items by the available shopper to the customer;
applying a hyperparameter learning model to the marketplace state data to predict a set of hyperparameters affecting a set of respective parameterized control decision models, wherein the hyperparameter learning model is trained on historical marketplace state data and a configured outcome objective for the online concierge system, wherein training the hyperparameter learning model comprises:
logging the marketplace state data over a period of time; and
re-training the hyperparameter learning model based on the logged marketplace state data;
independently applying the set of parameterized control decision models to the marketplace state data using the hyperparameters to generate a respective set of control parameters affecting marketplace operation of the online concierge system, comprising:
for each of the set of parameterized control decision models,
modifying parameters of the parameterized control decision model using the hyperparameters,
providing the marketplace state data as input to the modified parameterized control decision, and
receiving one of the set of control parameters as output from the modified parameterized control decision; and
applying the respective set of control parameters to modify an operation of the online concierge system.