Coloured Petri Nets Modeling Multilayer Perceptron Neural Networks

In critical sectors such as healthcare, where neural networks are increasingly applied, transparency and reliability in these systems are crucial. The present study explores the utilization of coloured Petri nets (CPN) to represent and analyze multi-layer perceptron (MLP) neural networks, introducing a structured methodology for evaluating critical systems that rely on MLP models for classification. A case study involving a COVID-19 dataset corroborates the efficacy of this approach by comparing the MLP and CPN models. This method simplifies the formal verification process and enhances the interpretability of machine learning systems.