Transforming IoT with Federated Multimodal Learning: A Privacy-Centric Approach

Suppliers face significant challenges when discussing cloud-based health apps, such as implementing highly secure procedures and rigorous privacy safeguards. This applies to the fact that the acceptance of these applications is limited by the need for such quality of service, which is achieved mainly by reducing latency. This work is devoted to tackling these hurdles through the cutting-edge application of edge computing and model-federated learning aspects. Federated learning, a complex and distributed paradigm for machine learning helps perform computations near the data source and overcomes the constraint problem of a distant data source. Integrating the newest edge computing, our system includes an intelligent instrument that is applied to extract clinical data using Clustered Federated Learning (CFL). The value of this framework is evaluated using the performance metrics of two benchmark datasets, one of which represents different kinds of problems that occur in the day-to-day life of people, which concludes the effectiveness and significance of implementing machine learning at the edge in the health care sector The presented results above strongly suggest the fact that CFL promotes this model results in improvement. The performance indicators of the CFL model showed an appraisal boost with the increments in X-ray and Ultrasound datasets by 16% and 11%, respectively, when compared to those trained conventionally. Hence, edge AI through CFL can achieve the same tasks, and the possible scenarios can also be remarkably superior compared to a centralized data system.