<p>Due to the rapidly growing number of internet users, security continues to play a crucial role in today’s online world. To detect and identify intruders, many researchers have developed a variety of intrusion detection techniques. In the meantime, the detection accuracy of the current technologies was not acceptable. Here, Namib Beetle Migration Optimization-based Deep Stacked Auto Encoder (NBMO_DSAE) is introduced for Intrusion Detection (ID). Initially, input data is processed. Then Z-score Normalization (ZN) is utilized for normalizing input data and thereafter, feature fusion (FF) is conducted by Deep Belief Network (DBN).) with Jeffrey similarity. After that, the oversampling approach is exploited to carry out data augmentation (DA). At last, the network ID is accomplished using a Deep Stacked Auto Encoder (DSAE), which is trained by NBMO. NBMO is an assimilation of Namib beetle optimization (NBO) with Wild Geese Migration Optimization (GMO). Furthermore, in UNSWNB15, NBMO_DSAE acquired accuracy of 0.952, precision of 0.931, Recall of 0.943 and F1-Score using 0.938. For CICIDS, NBMO_DSAE achieved accuracy of 0.941, precision of 0.921, Recall of 0.931 and F1-Score 0.925 This method is scalable and can handle large volumes of data, making it suitable for Network Intrusion Detection.</p>