Reversible Decimal First Degree Cellular Automata for Data Classification

The classification problem predicts labels for input data based on the training dataset. In this paper, the cyclic spaces of first degree reversible cellular automata are used to solve the classification problem. Every dataset of classification problem contains different classes. Based on the class labels in each configuration of cycles of the first degree reversible cellular automaton, they are grouped into different classes. The main advantage of this proposed model is that we can use real-world numerical data directly without doing any type of encoding. When selecting a reversible CA for classification, it is essential to maintain a minimum distance property within the same cycle’s configurations, while ensuring a significant distance between configurations from different cycles. This study identifies (linear) CAs that satisfy this criterion. Our model performs well in comparison to the existing machine learning models.