Amazon Technologies, Inc.

Jae In Cho

John Martin Miller

Kirtan Modi

Qie Hu Huang

Xinyu Zhou

Jiahui Jeff Wang

Xing Zhao

Techniques for utilizing machine learning to generate and use publishing parameters for a content delivery system are described. According to some examples, a computer-implemented method includes receiving, by a content delivery service, a request from a content provider device to send a media file to a client device; determining, by a provider intention match machine learning model of the content delivery service, a first set of one or more potential publishing parameters for the media file based on the request; sending a proposal to the content provider device to send the media file to the client device according to the first set of one or more potential publishing parameters; receiving, by the content delivery service, an indication from the content provider device to modify the first set of one or more potential publishing parameters for the media file; determining, by a negotiation simulation machine learning model of the content delivery service, a second set of one or more potential publishing parameters for the media file based on the indication from the content provider device; and sending the media file to the client device based on the second set of one or more potential publishing parameters.

Computer-implemented methods for determining publishing parameters for a content delivery system

Systems, devices, and methods are described herein for improving inventory management. As used herein, 'inventory' refers to digital space at an inventory providers webpage at which content can be delivered. The disclosed techniques utilize reinforced machine learning and an offline training process to train various models with which a content request corresponding to the inventory can be classified according to historical requests and a selection process identified for the request (e.g., a direct or an indirect selection process). If an indirect selection process is chosen, the content request may be optimized for that process utilizing additional machine learning models trained using reinforced machine learning and the offline training process. The disclosed techniques enable the inventory provider to optimize content selections according to a preferred objective. The training operations are performed offline, in a training system configured to simulate the run time environment.

Simulation augmented reinforcement learning for real-time content selection

Growing community of inventors

Hoboken, NJ, United States of America

Jae In Cho