User behavior model development with private federated learning

US12052315B2 · kind B2 · utility

Assignee

• Apple Inc. · US

Inventors

• Stephen Cosman · Redmond, US
• Kalu Onuka Kalu · Waterloo, CA
• Marcelo Lotif Araujo · Campbell, US
• Michael Chatzidakis · San Jose, US
• Thi Hai Van Do · Santa Clara, US
• Alexis Hugo Louis Durocher · San Francisco, US
• Guillaume Tartavel · Paris, FR
• Sowmya Gopalan · Cupertino, US
• Vignesh Jagadeesh · Cupertino, US
• Abhishek Bhowmick · San Francisco, US
• John Duchi · Menlo Park, US
• Julien Freudiger · Mountain View, US
• Gaurav Kapoor · Santa Clara, US
• Ryan M. Rogers · Los Gatos, US

Key dates

• Filing date: Dec 21, 2020
• Grant date: Jul 30, 2024
• Expiry date: Nov 2, 2042

Classification

• Technology area (CPC H): Electricity
• CPC primary: H04L67/306
• WIPO field: Computer technology
• WIPO sector: Electrical engineering

Abstract

Embodiments described herein provide for a non-transitory machine-readable medium storing instructions to cause one or more processors to receive, at a client device, a machine learning model from a server, detect a usage pattern for a content item, store an association between the content item and the detected usage pattern in local data, train the machine learning model using local data for the content item with the detected usage pattern to generate a trained machine learning model, generate an update for the machine learning model, privatize the update for the machine learning model, and transmit the privatized update for the machine learning model to the server.