This research explores a network intrusion detection method that combines feature selection with adoptive voting. This technique makes advantage of the NSL-KDD dataset, a popular high-dimensional dataset for detecting network intrusions. This was accomplished by an employ of a huge range of data preparation techniques, such as shape validation, concatenation, numeric conversion, feature selection, and data balancing. In the subsequent step, a dataset was splitted into training and testing sets using a conventional split ratio of 80:20. A significant AUC value that was close to 0.998 was seen in the results of the evaluation of the Convolutional Neural Network (CNN), which demonstrated high levels of accuracy, precision, recall, and F1-scores ranging from 0.963 to 0.986. It is possible that future work will involve investigating ensemble approaches or expanding the study to include real-time IDSs. An objective of this work is to further improve the resilience and application of algorithms.