Results of the research of algorithm creating capabilities using convolutional neural network

Abstract

The current technological momentum indicates that demand will soon rise for programming skills. For this reason, finding young individuals with these skills and orienting them towards programming related career paths will be highly important. In our research, we developed an age and programming knowledge independent system, which can help to identify talented youngsters. Before filling the test, we collected information through an anonymous survey about the test subjects’ IT knowledge, which can help identify further correlations at the time of result evaluation. First we had to create a measurement system which will provide objective test results about the children’s algorithmic thinking. We developed a web application that is responsive with a camera which follows the client eye moving and gather data into our database. This data is then passed into a machine learning algorithm to create a model and from it we could read our conclusions. The main goal of machine learning using convolutional neural network is predicting how the algorithmic thinking will develop in the case of children. We used a more game-like approach because it is more natural to children to think when they think they are playing some game. The game is a simple but effective LED lightning game and is very similar to a memory-based game.