A naukri.com initiative
Arxiv
4d
16
Image Credit: Arxiv
Redefining Clustered Federated Learning for System Identification: The Path of ClusterCraft
- The paper introduces an algorithm called IC-SYSID for addressing the System Identification (SYSID) problem in federated learning without prior dataset knowledge.
- IC-SYSID uses an incremental clustering method called ClusterCraft (CC) to group similar local workers together and reduce dependence on prior dataset knowledge.
- To enhance IC-SYSID, methods like ClusterMerge, enhanced ClusterCraft, regularization in loss function, and initializing cluster models are utilized.
- Experiments with IC-SYSID on a real-world SYSID problem of collaborative learning in vehicle dynamics demonstrate high performance and stability in cluster models.
Read Full Article
Like
For uninterrupted reading, download the app