• P Vandana
• Mehrunnissa

Abstract: With its track record of success in a variety of fields, including speech recognition and image analysis, deep learning has garnered a lot of attention. Deep learning is very effective at handling high-dimensional data that exhibits nonlinearity. Our initial analysis of network traffic data indicates a very high degree of non-linearity, which helps to explain why applying standard machine learning methods (e.g., SVM, Random Forest, Adaboosting) to increase detection accuracy is difficult. To determine whether deep learning is feasible for network anomaly detection, we empirically assess it in this work. Using two public traffic data sets, we analyse a collection of deep learning models built on the Fully Connected Network (FCN), Variational AutoEncoder (VAE), and Long Short-Term Memory with Sequence to Sequence (LSTM Seq2Seq) architectures. that, when it comes to the distribution of normal and attack populations, have unique characteristics. With 99% binary classification accuracy on public data sets, the model based on the LSTM Seq2Seq structure demonstrates a highly promising performance, and our experimental results validate the promise of deep learning models for network anomaly detection.

Keywords: Network anomaly detection, Deep learning, LSTM, Seq2Seq, the Fully Connected Network (FCN), Variational AutoEncoder (VAE)

"Deep Learning: An Empirical Assessment for Network Anomaly Detection", International Journal of Emerging Technologies and Innovative Research (www.jetir.org), ISSN:2349-5162, Vol.11, Issue 2, page no.d37-d41, February-2024, Available :http://www.jetir.org/papers/JETIR2402305.pdf

"Deep Learning: An Empirical Assessment for Network Anomaly Detection", International Journal of Emerging Technologies and Innovative Research (www.jetir.org | UGC and issn Approved), ISSN:2349-5162, Vol.11, Issue 2, page no. ppd37-d41, February-2024, Available at : http://www.jetir.org/papers/JETIR2402305.pdf