**2. Materials and Method**

This section presents the characteristics of the dataset used, as well as the GDL technique discussed in this work. Naturally, to apply a convolutional neural network on graphs, additional pre- and post-processing had to be performed. The whole processing flow is shown in Figure 1; however, the majority of the details of the methods used will be presented in sections devoted to the specific experiments. Here, only the functioning of the neural network will be explained. Moreover, while describing the details of the GDL approach, the hypothesis about its properties will be formulated. That is why, at the end of this section, the results of the experiment, verifying the correctness of the proposed assumptions, are described as well.
