Multi-Stage Federated Learning in Wireless Networks

ChangeBridge: Patent Apps - AI & Computing (G06N)

Published April 2nd, 2026

Detected April 3rd, 2026

Summary

USPTO published patent application US20260094032A1 for multi-stage federated learning in wireless networks, filed September 15, 2023. Inventors: Ping-Heng KUO and Alexander SIROTKIN. The application covers a method where AI agents train partial models, which are aggregated into a global model by an AI manager, with a trustworthiness determination gate before proceeding to the next stage.

View original document View source feed page

What changed

USPTO published patent application US20260094032A1 (Application No. 19113252) on April 2, 2026, covering a multi-stage federated learning system for wireless networks. The invention describes a method where a first group of AI agents train and report partial AI/ML models to an AI manager, which aggregates them into a first version of a global model. If the first version is determined trustworthy, a second group of AI agents trains additional partial models that are aggregated with the first version to produce a second, improved global model.

This patent application does not impose compliance obligations. Technology companies developing federated learning systems for wireless networks, AI researchers, and telecommunications equipment manufacturers should review the published claims for potential freedom-to-operate considerations. The patent is classified under CPC G06N 5/043 and has a filing date of September 15, 2023.

Source document (simplified)

← USPTO Patent Applications

Multi-Stage Federated Learning in Wireless Networks

Application US20260094032A1 Kind: A1 Apr 02, 2026

Inventors

Ping-Heng KUO, Alexander SIROTKIN

Abstract

first group of AI agents to train and report, per each AI agent of the first group, a respective first partial AI or machine learning (ML) (AI/ML) model to the AI manager, receive the first partial model from each AI agent of the first group, generate a first version of a global model from the first partial models, if the first version of the global model is determined to be trustworthy, select a second group of AI agents to train and report, per each AI agent of the second group, a respective second partial AI/ML model to the AI manager, receive the second partial models and aggregate the second partial models and the first version of the global model into a second version of the global model.