Synthetic networks for classification methods

<p>The current study presents a collection of synthetic network databases intended for evaluating classification methods. The first database, termed &quot;4-models,&quot; comprises synthetic networks that have been generated based on four distinct network classes, namely, Random, Small-world, Scale-free, and Geographical. This database includes 2800 networks for each class, resulting in a total of 11,200 networks. The second database, named &quot;Scale-free,&quot; is composed of scale-free synthetic models generated using the models proposed by Barab&aacute;si and Albert and by Dorogovstev and Mendes. This database includes five classes, comprising four Barabasi models based on the network&#39;s power law parameter and a Mendes model, each class consisting of 100 networks. The third database, named &quot;Noisy,&quot; is created by modifying the topology of the networks from the 4-models and Scale-free databases via the addition or removal of edges. This database contains three sub-datasets, based on the level of randomness added to the networks, and eight different classes, each of which includes 100 networks.</p> <p>To ensure consistency and comparability of our results, we implemented a standardization procedure whereby all databases were converted to adjacency list format. This allowed us to more efficiently process and analyze the data.</p>