NodalNet: A Node Based Programming Interactive Educational Tool in Developing Neural Network for Beginners

Machine Learning, a subset of Artificial Intelligence, is now being integrated into image and speech classification, object and action recognition, medical image analysis, security and surveillance, and autonomous driving. It is usually executed by creating a neural network which is defined as the process of modeling a computer system to mimic the function of the human brain. However, learning to develop a neural network is a challenge for students, especially those who have little background in programming. This study focuses on developing a node-based programming tool called NodalNet, a beginner-friendly and interactive educational tool for creating a neural network designed for image classification applications. Java programming language, Java swing library, and Eclipse integrated development environment (IDE) are used to build the NodalNet. It can be used for offline deployment using laptops, and computers, and for online deployment using a technique called the application programming interface (API) method. NodalNet has built-in components that can be connected through nodes, and it has a data labeling feature that helps the users in classifying the raw data inputs used before the training process. The NodalNet can develop and train simple neural networks, some image recognition samples are FlowerNET, DogCatNET, and NumberNET which have significant results in their accuracy. A survey conducted shows that 77.78% of the respondents answered that it is easy to train using the NodalNet. This tool offers a positive impact as an interactive educational tool for developing a neural network.