The Public are considerably using the various types of online social networks (OSNs) and it is become more common in people's social life. Thus, the users are facing spam relate issues and fake accounts due to Out-of-controlOSNs evolution, due to these attacks users personal information is remains unsafe . To solve these problems , various types of machine learning algorithms are propo sed by the various Researchers.But these methods are failed to detect the bots, spam detection and fake accounts detection effectively with maximum accuracy. Thus, this paper proposes to use the Deep Learning Convolutional Neural Network ( DLCNN)as a modern algorithm to effectively identify suspected ClickstreamSequences and bots, to add choices and to restrict measurements. Th e classification mastering algorithmis used to determine the act ual or false identity of target fake accounts . From the extensive simulation results, it is observed that the proposed DLCNN consumes less training time and provides highest classification accuracy compared to the state of art approaches.