Analysis of Forest Fire Data Using Neural Network Rule Extraction with Human Understandable Rules

Forest fires spread fast, uncontrollably, and may lead to massive destruction. This makes the prevention of them a safety critical issue. Neural networks are a sub-area of machine learning that can be used to analyze the complex behavior of neural systems and help to predict forest fires. To make the knowledge learned by a neural network more accessible, rules can be extracted from the neural network to demystify the system behavior and directly relate inputs to outputs. In this paper, we present a Dynamic Cell Structure (DCS) neural network used for forest fire data prediction, determining which environmental factors lead to fires. We apply an intuitive rule extraction algorithm to extract understandable rules for this prediction. The results are verified through direct comparison with the raw data.