| Summary: | 碩士 === 國立中正大學 === 資訊管理所 === 96 === In recent years, the internet and the pc become more and more widespread. The internet-based services are widely adapted by enterprises and governments. And information security plays a more and more important role in these organizations. When one organization suffers internet attack, the loss is huge. Organizations often use intrusion detection System（IDS） to stop and prevent internet attacks from crackers. And many intrusion detection methods were proposed in these years, such as machine learning methods. How about the performance of these methods? So we want to compare the performance of machine learning methods in intrusion detection systems. In this paper, we compare the performance of decision tree and support vector machine. We use the benchmark dataset, KDD Cup 99 dataset. We compare the accuracy, detection rate, false alarm rate, and accuracy of the four classes of attack, including Probe, Dos, U2R, R2L. Finally, some suggestions are proposed for the two methods.
|
|---|
