The following machine learning methods were employed in the four experiments that were discussed in this study: DT, RF, GB, MLP, and Making a Vote: In the first trial, DT, MLP, and RF were utilized to detect network attacks; in the second trial, RF, DT, and XGB were employed to detect hostile attacks. As a first step in this research, we examine the UNSW NB15 dataset, which details several forms of network attacks. Second, we got the dataset ready to feed into machine learning methods so we could produce many models. Included in this were activities such as using user min_max_normalize to guarantee that all datasets fell within the same ranges and applying a Label Encoder to transform non-numerical attributes into numerical ones. The machine learning algorithms (ML) results were tested according to the previously mentioned standards. Compared to other detection approaches, DT and the voting strategy performed better. The results show that MLs are quite good at detecting and differentiating between malicious and legitimate attacks.