Federated Learning with Backbone Decoder Models

ChangeBridge: Patent Apps - AI & Computing (G06N)

Published March 26th, 2026

Detected March 31st, 2026

Summary

USPTO published patent application US20260087416A1 by Sony Group Corporation covering apparatus and methods for federated learning using a backbone-decoder model architecture. The invention enables servers to distribute decoder components to edge devices for localized training while maintaining backbone model parameters centrally, with aggregated decoder updates returned to the server for model refinement.

View original document View source feed page

What changed

Sony Group Corporation filed USPTO Application US20260087416A1 disclosing a federated learning system where processing circuitry generates a second machine-learning model comprising a backbone and decoder from a first model. The server distributes the decoder to devices (and the backbone for the first iteration), receives trained decoder updates from devices, and iteratively refines the decoder before updating the first model's decoder.

This patent application represents intellectual property filing activity rather than a regulatory requirement. Companies developing federated learning systems should monitor this published application to understand Sony's claimed scope for potential freedom-to-operate considerations or licensing discussions. No compliance actions are required as this is an informational publication of a patent application, not an issued patent or regulatory mandate.

Source document (simplified)

← USPTO Patent Applications

APPARATUS AND METHODS FOR FEDERATED LEARNING, DEVICE AND METHOD FOR A DEVICE

Application US20260087416A1 Kind: A1 Mar 26, 2026

Assignee

Sony Group Corporation

Inventors

Weiming ZHUANG, Jingtao LI, Lingjuan LYU

Abstract

The federated learning of a first machine-learning model apparatus includes processing circuitry configured to generate a second machine-learning model including a backbone and a decoder from the first machine-learning model. The processing circuitry is configured to perform at least one iteration of the following: (a) output the decoder of the second machine-learning model to one or more devices and for the first iteration of the at least one iteration further output the backbone of the second machine-learning model to one or more devices; (b) receive a trained version of a decoder for the second machine-learning model from one or more devices; and (c) update the decoder of the second machine-learning model based on the trained version of the decoder received from one or more devices. The processing circuitry is configured to update a decoder of the first machine-learning model based on the updated decoder of the second machine-learning model.