Optimized Feed Forward Neural Network for Fake and Clone Account Detection in Online Social Networks

Online Social Network (OSN) is becoming popular among peoples to communicate, share and disseminate the personal, social and technical information among the other user in the Social media. Despite of its popularity, many challenges has been exhibiting on usage of the network in terms of security breaches. Especially fake account and clone attack has been propagating in the network on obtaining or misusing of others information’s which leads to privacy threats to the user of the account. In order to mitigate and eliminate those issues, machine learning based technique is employed to detect and prevent the fake account and clone attack propagation in the network. In this paper, a new optimized feed forward neural network is employed to detect the fake account. Initially data is preprocessed to remove the missing values and non-available values. In this, outlier data of the data stream is removed by employing the K-means clustering, it reduces the training time of the user classification and prediction process. The prediction of fake account in the data streams of the social network data is carried out using neural network learning model mentioned as feed forward neural network. The neural network considers the inputs on the data streams as identifying the dominant attributes on employing the principle component analysis for feature extraction process and genetic algorithm for optimal feature selection. The objective function of the neural network is optimized to reduce the misclassification rate of the user classifier. This proposed approach is able to enhance the accuracy of the user classification on comparing with existing technique on its evaluation with twitter dataset. The whole process is realized and evaluated with respect to accuracy, precision and recall.