This study investigates the performance of intrusion detection systems (IDS) using the UNSW-NB15 dataset, focusing on enhancing detection accuracy and reducing false positives. The research employs four machine learning algorithms—logistic regression, support vector machine (SVM), decision tree, and random forest—to analyze network traffic data and identify cyber-attacks. Through in-depth exploratory data analysis (EDA), feature selection, and model training, the study evaluates the effectiveness of these algorithms in detecting cyber-attacks. The results show that the random forest model outperforms the others, achieving an F1 score of 97.80%, accuracy of 98.63%, and a low false alarm rate of 1.36%. The study also highlights the importance of feature selection in improving model performance and reducing false positives. The UNSW-NB15 dataset, which contains 2,540,044 network traffic records, is used to train and test the models. The dataset includes various types of cyber-attacks, such as fuzzers, backdoors, exploits, and worms. The study concludes that the random forest model is the most effective for detecting cyber-attacks, offering high accuracy and low false alarm rates. The research also discusses the challenges of data imbalance and the need for further optimization in future work. The findings demonstrate the potential of machine learning in enhancing the performance of intrusion detection systems and improving network security.This study investigates the performance of intrusion detection systems (IDS) using the UNSW-NB15 dataset, focusing on enhancing detection accuracy and reducing false positives. The research employs four machine learning algorithms—logistic regression, support vector machine (SVM), decision tree, and random forest—to analyze network traffic data and identify cyber-attacks. Through in-depth exploratory data analysis (EDA), feature selection, and model training, the study evaluates the effectiveness of these algorithms in detecting cyber-attacks. The results show that the random forest model outperforms the others, achieving an F1 score of 97.80%, accuracy of 98.63%, and a low false alarm rate of 1.36%. The study also highlights the importance of feature selection in improving model performance and reducing false positives. The UNSW-NB15 dataset, which contains 2,540,044 network traffic records, is used to train and test the models. The dataset includes various types of cyber-attacks, such as fuzzers, backdoors, exploits, and worms. The study concludes that the random forest model is the most effective for detecting cyber-attacks, offering high accuracy and low false alarm rates. The research also discusses the challenges of data imbalance and the need for further optimization in future work. The findings demonstrate the potential of machine learning in enhancing the performance of intrusion detection systems and improving network security.