Title:

Statistical Analysis for Applications of Federated Machine Learning using House Plants

This project aims to demonstrate the capabilities used for Federated Machine Learning (Fed-ML) which can be used to provide a sense of both security and efficiency for IoT applications. The motivation stems from the absence of prior research leveraging Fed-ML in the context of houseplant monitoring and the necessity to develop a robust AI/ML predictor utilizing a multi-sensor, multi-node framework. The applicability assessment involved statistical analysis of three Fed-ML systems: Cross-Device, Cross-Silo, & Split. The infrastructure setup required to enable this project was done by utilizing IoT sensor nodes (Raspberry Pi 4 & Raspberry Pi Zero W) along with edge devices (Raspberry Pi 4) to allow for the Fed-ML system to operate. This study seeks to investigate the usability and tradeoffs associated with Fed-ML and the approaches necessary to improve the usability of this system. With a multi-sensor & multi-nodal system, the scalability for this project can improve the quality of the results. The outcomes of this study shows that the application of utilizing this format of machine learning can be beneficial for a variety of use cases.

Name

Comment*

Comments are viewable only by submitter

Captcha*