Optic Clustering

Optic clustering is a technique used to improve the performance of federated learning, a method for training machine learning models on decentralized data while preserving privacy. Current research focuses on adapting clustering algorithms, such as OPTICS, to dynamically optimize the aggregation of locally trained models from diverse data sources, addressing the challenge of non-identical data distributions across devices. This adaptive approach aims to enhance the accuracy and efficiency of federated learning by intelligently grouping similar data before model aggregation. The resulting improvements have significant implications for various applications requiring distributed data analysis while maintaining data privacy.

Papers

June 22, 2023

Reinforcement Federated Learning Method Based on Adaptive OPTICS Clustering
Tianyu Zhao, Junping Du, Yingxia Shao, Zeli Guan
Federated Reinforcement Learning Optic Clustering

Optic Clustering

Papers

Reinforcement Federated Learning Method Based on Adaptive OPTICS Clustering