DYNAMIC FEDERATED LEARNING FRAMEWORK FOR PRIVACY-CENTRIC AI IN CLOUD ECOSYSTEMS

Abstract

The present invention relates to a dynamic federated learning framework designed to enable privacy-preserving artificial intelligence (AI) in cloud ecosystems. The framework includes a central server responsible for coordinating the federated learning process, dynamically selecting federated nodes for participation, and securely aggregating model updates. Each federated node trains machine learning models locally on private data, sending only model updates to the central server, ensuring data privacy. The system features a dynamic participant selection mechanism that adapts in real-time based on data heterogeneity, computational resources, and privacy requirements. The aggregation strategy is adaptive, accounting for data quality and node capabilities. Privacy is safeguarded using secure multi-party computation (SMC) and differential privacy techniques. Real-time monitoring ensures fault tolerance and robust performance by detecting and addressing issues such as faulty participants or network failures. This f

Specification

Description:The embodiments of the present invention generally relate to a dynamic federated learning framework that enables privacy-preserving artificial intelligence (AI) in cloud ecosystems. More specifically, it concerns systems and methods for securely and efficiently managing federated learning in a decentralized environment, where data privacy, security, and model performance are key considerations. This invention is particularly useful in cloud-based infrastructures that require robust, scalable AI models, while ensuring compliance with privacy laws and protecting sensitive user data.
BACKGROUND OF THE INVENTION
The following description of related art is intended to provide background information pertaining to the field of the disclosure. This section may include certain aspects of the art that may be related to various features of the present disclosure. However, it should be appreciated that this section be used only to enhance the understanding of the reader with respect to the present disclosure, and not as , Claims:1. A dynamic federated learning framework for privacy-preserving artificial intelligence (AI) in cloud ecosystems, comprising:
a central server configured to coordinate the federated learning process, dynamically select federated nodes for participation, aggregate model updates from the selected federated nodes, and ensure data privacy;
a plurality of federated nodes, each configured to locally train a machine learning model on private data without transmitting raw data to the central server, wherein each federated node sends only model updates to the central server;
a dynamic participant selection mechanism implemented by the central server, wherein participants are selected based on real-time criteria, including data heterogeneity, computational resources, and privacy requirements;
an adaptive model aggregation module configured to aggregate the model updates from the federated nodes using an aggregation strategy that accounts for data quality, data diversity, and computational capabilities of the federated

Documents