Network Intrusion Detection System using ML

Abstract

Technologies are making our life easier and simple, but it has both positive and negative effect. Many new methods of cybercrimes are coming which cannot be solved using earlier conventional method like using firewalls, antivirus, old ML algorithms. In recent years every device whether it’s hardware or software is being connected with IOT. Hence, there is huge growth in data also their privacy is huge concern for industries. In this model, we are implementing Network Intrusion detection system using Machine learning algorithms which would resolve security problems using KNN, SVM, LR, RF, DT and Gaussian NB with greater efficiency. Our system uses both supervised and unsupervised machine learning techniques. Both misuse and Anomaly based detection to detect malware and viruses, our system is capable to detect both known and unknown attacks. In case of misuse detection system known attacks are being easily identified using a database where list of all known attacks is available. If any attack happens on network system, then NIDS checks whether the attack is listed in dataset or not. If attack is known, then system administrator gets notified. If attack is unknown, then NIDS uses outlier detection to identify attack using several machine learning algorithms like clustering and other techniques. So, with the help of above-mentioned techniques attack is being detected. Our model improves the attack detection mechanism with high accuracy and less prediction time. It is better than previous conventional machine learning algorithms. Our model is broadly accepted in companies and organization. it is fulfilling the cyber security issue also threat prediction time of our model is quite improved and the prediction time is reduced as compared to previous model.