| Autor: |
B. Vishnupriya, R. S. Soundariya, M. Nivaashini, R. M. Tharsanee, G. Pavithra |
| Rok vydání: |
2021 |
| Předmět: |
|
| Zdroj: |
Lecture Notes in Networks and Systems ISBN: 9783030847593 |
| Popis: |
Cyber-attacks have become highly critical in recent times and impose more difficulty in implementing appropriate intrusion detection systems. Security principles in a system would be at risk if the intrusions are not detected accurately. Several state-of-the-art techniques for Intrusion detection proposed earlier are mainly categorized under the Misuse based and Anomaly based methods. This paper implies Nature Bio-Inspired algorithms and Machine learning algorithms for Intrusion Detection. The distribution of attacks in the training dataset is highly skewed in the KDDCup 99 dataset. Feature Selection is a challenging task on massive data, selecting an optimized subset of features will often lead to better performance. In this work, Binary-Firefly Algorithm is used as the feature selection technique. This algorithm is implemented in a Hadoop environment using MapReduce Programming. After the feature selection process, the dataset is then trained and tested using cascaded classifiers (Naive Bayes Classifiers and J48). The results obtained using cascaded classifiers are compared against the results produced by single classifier and parallel classifier. The performance comparison shows that the proposed system identifies the most optimal feature set with improved detection rate, compared to existing methods. |
| Databáze: |
OpenAIRE |
| Externí odkaz: |
|
