 Our proposed deep learning-based approach using self-taught learning, STL, on NSLKDD dataset presents an efficient and flexible network intrusion detection system, NIDS, for unforeseen and unpredictable attacks with better performance compared to previous work in terms of accuracy, precision, recall, and F-measure values. This article was authored by Amad Javade, Quamar Niyaz, Waching Sun, and others.