US 11,755,906 B2
System for dynamic estimated time of arrival predictive updates
Jeff Ning Han, Palo Alto, CA (US); William Preston Parry, Akron, OH (US); Bing Wang, Palo Alto, CA (US); and Rohan Balraj Chopra, San Francisco, CA (US)
Assigned to DoorDash, Inc., San Francisco, CA (US)
Filed by DoorDash, Inc., San Francisco, CA (US)
Filed on May 7, 2021, as Appl. No. 17/314,487.
Application 17/314,487 is a continuation of application No. 15/798,207, filed on Oct. 30, 2017, granted, now 11,037,055.
Prior Publication US 2021/0264275 A1, Aug. 26, 2021
This patent is subject to a terminal disclaimer.
Int. Cl. G06N 3/08 (2023.01); G06Q 10/0833 (2023.01); G06Q 10/083 (2023.01); G06F 16/487 (2019.01); G06F 16/48 (2019.01); G06Q 50/12 (2012.01)
CPC G06N 3/08 (2013.01) [G06F 16/487 (2019.01); G06F 16/489 (2019.01); G06Q 10/0833 (2013.01); G06Q 10/0838 (2013.01); G06Q 50/12 (2013.01)] 20 Claims
OG exemplary drawing
 
1. A server comprising a processor and memory, wherein the processor is configured to:
train a neural network using a training dataset to dynamically output estimated time of arrival (ETA) predictive updates, wherein the neural network includes a plurality of subnetworks, each subnetwork corresponding to a duration between two successive events of a series of events for an order; and
receive a first confirmation message from a user device, the first confirmation message including first information corresponding to a first event of the series of events; and
input the first information into the neural network to automatically generate a first ETA prediction using weighted factors in the neural network, the first ETA prediction corresponding to the duration for a target event of the series of events, wherein the first ETA prediction is based on at least one output of the plurality of subnetworks.