• V V Nagamani

Machine learning-based approaches to IDS are designed to detect attacks, which rely on features extracted from network traffic data. The problem is to form attack signatures in terms of those features correlated to attack packets which is not possible for zero day attacks as there are very limited number of instances. Here a feature based transfer learning framework and an approach is introduced to overcome the variants of attacks causing the drop of detection performance. We extract the features from the raw data and a common latent feature representation is learned from both source and target domain via feature-based transfer learning algorithms. We project the source data and the target data into common k dimensional latent space with the help of the algorithm and then this representation is fed into the common base classifier to achieve better accuracy of detecting the unknown attacks.

Machine Learning, Transfer Learning, Network Attack Detection

"Heterogeneous Transfer Learning Algorithm for Network Security", International Journal of Emerging Technologies and Innovative Research (www.jetir.org), ISSN:2349-5162, Vol.6, Issue 6, page no.416-421, June 2019, Available :http://www.jetir.org/papers/JETIR1907H65.pdf

"Heterogeneous Transfer Learning Algorithm for Network Security", International Journal of Emerging Technologies and Innovative Research (www.jetir.org | UGC and issn Approved), ISSN:2349-5162, Vol.6, Issue 6, page no. pp416-421, June 2019, Available at : http://www.jetir.org/papers/JETIR1907H65.pdf