Efficient Deep CNN-BiLSTM Model for Network Intrusion Detection3rd International Conference on Artificial Intelligence and Pattern Recognition @ Huaqiao University, Xiamen, China (25th - 27th Sept 2020)
The model proposes an effective approach towards Network Intrusion Detection in cloud systems with active monitoring using Neural Networks. The proposed model combines spatial and temporal feature sets from CNN and BiLSTM on time range datasets to accurately predict network attacks and the category of attack to which they belong to. - PreprintCode
PhishX: An Empirical Approach to Phishing DetectionPreprint available
Proposed a new dataset with 198 features and provided detailed analysis on effectively detecting phishing websites by training models (Classifiers and Neural Networks) on a combination of 73,000 phishing websites with 52,000 top Alexa websites. Random Forest gives an accuracy of 93.09% with a FPR of 0.1222! - PreprintCode