As technological capabilities continue to expand at an exponential rate, our reliance on the internet for executing daily activities intensifies, rendering us increasingly susceptible to cyber threats. This augmented dependence on online platforms amplifies our exposure to numerous security vulnerabilities. Consequently, the demand for robust antivirus software or analogous systems capable of identifying and mitigating emerging security risks is escalating. An efficacious solution to this predicament is a system, either hardware or software-based, that can be deployed within a network infrastructure to detect and prevent various forms of intrusions, commonly referred to as an Intrusion Detection System (IDS). This system is instrumental in monitoring network traffic and discerning any anomalies. This study examines an IDS utilizing machine learning and neural networks (ML-NN), which exhibits superior classification accuracy across diverse attack types within the CIC-IDS dataset, surpassing current methodologies. The experimental analysis encompasses both binary and multiclass data classifications, employing optimal evaluation metrics—precision, accuracy, and recall—for the ML-NN approach, considering the specific nature of attack types and suggesting avenues for further investigation. The suggested framework achieves a precision of 99.87%, a recall of 99.89%, and an accuracy of 99.0%, which is 1.07% higher than previous approaches.