Intrusion Detection Using Deep Learning
Misbah Anwer, Ghufran Ahmed, Adnan Akhunzada, Shahbaz Siddiqui · 2021 International Conference on Electrical, Computer, Communications and Mechatronics Engineering (ICECCME) · 2021
Advancement in network attacks requires strong and growing security mechanisms. Internet of Things (IoT) is the evolving technology connected billion of devices right now and it builds on a set of network devices therefore it is under serious threat concern. Identifying attacks is crucial and critical task. The authors propose a hybrid DL driven approach to detect the attacks, one is Cuda Deep Neural Network Long Short-Term Memory (CuDNNLSTM) and another is Long Short-Term Memory (LSTM) on kitsune dataset. There is alarming situation for protection of all the smart systems in terms of security. In this paper we implemented LSTM and cuDNN LSTM networks to identify attack. Results show that our technique cuDNNLSTM outperforms in comparison of deep learning technique LSTMs that shows 99.79% accuracy on 6GB dataset approx. (2S0lac) records.