Discussion of the master student (Gheed Tawfiq Walid) in the Department of Computer Science / University of Technology on her thesis entitled (Improve Unsupervised Deep Neural Network to Enhance Anomaly Detection in Secure Mobile Network) This thesis proposes an improved model for detecting fraud cases of Banking credit card transactions. The proposed model consists of three stages, the first stage processes and reduces the dimensions of the data set and prepares it for the next stage, which is the unsupervised training phase. At this stage, the learning process is carried out through one of the concepts of deep learning, which is the automatic stack encoding, which is represented by the use of three layers of the neural network trained using different activation functions, which strive to generate close data for each of the classes that are passed through these three layers. This stage is followed by a group of operations, the first of which is to improve the results using Adam algorithm followed by the process of calculating the difference between the results and the data entered to reduce the error rate and then the process of classification of data to fraud or normal .. And that was on Thursday, (26/12/2019) at the Discussions hall in the department. The discussion committee was composed of (Prof. Dr. Ahmed Tariq Sadiq and Prof. Dr. Hanaa' Mohsen Ahmed) from the University of Technology / Department of Computer Science and (Dr.Abeer Salem Jamil) from Al Mansour University College / Department of Computer Science in the presence of the supervisors of the student (Dr. Abeer Tariq Mawloud and Prof. Dr. Abdul Mohsen Jaber Abdel Hussein).

Download Abstract