Transfer Learning Based Automatic Model Creation Tool For Resource Constraint Devices
This tool aims to simplify the creation of custom ML models for resource-constrained devices for developers, addressing the challenge of deploying ML on limited hardware.
This paper proposes an automatic model creation tool that leverages transfer learning to generate custom machine learning models for resource-constrained devices without requiring users to write ML code. The tool was used to create image and audio classifiers, demonstrating its ability to produce models with reported accuracy and memory footprint on the Stanford Cars and ESC-50 datasets.
With the enhancement of Machine Learning, many tools are being designed to assist developers to easily create their Machine Learning models. In this paper, we propose a novel method for auto creation of such custom models for constraint devices using transfer learning without the need to write any machine learning code. We share the architecture of our automatic model creation tool and the CNN Model created by it using pretrained models such as YAMNet and MobileNetV2 as feature extractors. Finally, we demonstrate accuracy and memory footprint of the model created from the tool by creating an Automatic Image and Audio classifier and report the results of our experiments using Stanford Cars and ESC-50 dataset.