Design and Implementation of Educational Neural Networks Visual Interface Program

Abstract

In this study, a visual computer application was carried out in order to be a support material for the students studying on Artificial Neural Networks, which were developed inspired by the operation of the nerve cells of living things and used in solving nonlinear problems such as classification or regression. CSharp.Net, an object-oriented software development tool, has been used for easy management of the application in which complex algorithms will be executed due to its structure. The most important purpose of the implemented software and its advantage over other similar applications is that it is user-friendly, that is, easy to use. The student who has studied the subject theoretically will be able to easily design the artificial neural network, the diagram of which can be followed visually, train the network and instantly monitor the minimum error graph reached at the end of the training. While performing these processes, an application has been designed in which the student can manage the application with the skills of the mouse, without the need to know any programming language, that is, without writing any code, even using the keyboard at a minimum. Since the main purpose of my application is to comprehend the subject with the help of visual elements, it is focused on a single problem solution (regression). The data of the logical logic gates, which are widely used in teaching the subject, are given ready in the application interface, and for other problem scenarios, an external data entry procedure is provided from the Excel application in order to easily enter the data set into the application.