Optimalizace metody strojového učení pro aplikace v mikrokontrolérech

Abstract

This thesis addresses the optimization of machine learning for implementation on microcontrollers, specifically on the NXP i.MX RT1170 platform. The main goal of the work is to assess the feasibility and efficiency of deploying artificial neural networks (ANN) in embedded systems. The study explores the use of the NXP eIQ software tool, which is designed to support and optimize ANN on microcontroller platforms. I analyze various optimization methods that affect deffrent model parameters, such as model size, prediction accuracy, and computational demand.