17766955. MODERATOR FOR FEDERATED LEARNING simplified abstract (Telefonaktiebolaget LM Ericsson (publ))

MODERATOR FOR FEDERATED LEARNING

Organization Name

Telefonaktiebolaget LM Ericsson (publ)

Inventor(s)

Swarup Kumar Mohalik of Marathahalli, Bangalore (IN)

Perepu Satheesh Kumar of Velachery, Chennai (IN)

Anshu Shukla of Bangalore (IN)

MODERATOR FOR FEDERATED LEARNING - A simplified explanation of the abstract

This abstract first appeared for US patent application 17766955 titled 'MODERATOR FOR FEDERATED LEARNING

Simplified Explanation

The abstract describes a method for training a central model in a federated learning system. Here is a simplified explanation of the abstract:

• The method involves receiving updates from multiple local models in a set of local models.
• These updates are enqueued in one or more queues corresponding to the local models.
• An update is selected from the queues based on a selection criteria related to the quality of the central model.
• The selected update is then applied to the central model or instructed to be applied by a node.

Potential Applications:

• Federated learning systems: This method can be applied in various domains where federated learning is used, such as healthcare, finance, and IoT, to train a central model using updates from multiple local models.

Problems Solved:

• Efficient training: By enqueuing and selecting updates from multiple local models, this method allows for more efficient training of a central model in a federated learning system.

Benefits:

• Improved model quality: The selection criteria based on the quality of the central model ensures that only the most relevant updates are applied, leading to an improved overall model quality.
• Scalability: The method can handle updates from a large number of local models, making it scalable for federated learning systems with a large number of participants.
• Decentralized training: By allowing local models to contribute updates, the method enables decentralized training while maintaining the integrity of the central model.

Original Abstract Submitted

A method for training a central model in a federated learning system is provide. The method includes receiving a first update from a first local model of a set of local models; receiving a second update from a second local model of the set of local models; enqueueing the first update and the second update in one more queues corresponding to the set of local models; selecting an update from the one or more queues to apply to a central model based on determining that a selection criteria is satisfied, the selection criteria being related to a quality of the central model; and applying the selected update to the central model or instructing a node to apply the selected update to the central model.