Fake Detection in Imbalance Dataset by Semi-Supervised Learning with GAN

As social media grows faster, harassment becomes more prevalent, which leads to the consideration of fake detection as a fascinating field among researchers. The graph nature of data, with a large number of nodes, causes different obstacles, including a considerable amount of unrelated features in matrices, as well as high dispersion and imbalanced classes in the dataset. To address these issues, Auto-encoders and a combination of semi-supervised learning and the GAN algorithm, called SGAN, were used. This paper deploys a smaller number of labels and applies SGAN as a classifier. The results of this test showed that the accuracy reached 81% in detecting fake accounts using only 100 labeled samples.