I am a student working on my thesis about network intrusion detection. I need to build a pure 1D CNN model (no hybrids like CNN-LSTM) to use as a baseline for comparison on the NSL-KDD and CIC-IDS2017 datasets. I have seen many examples, but most are for other domains (like IMU sensors) or are complex hybrid models. I am looking for a standard, reliable implementation that includes: The correct way to preprocess these specific datasets for a 1D CNN. A simple 1D CNN architecture (e.g., with Conv1D, pooling, and dense layers). The right input shape (samples, timesteps, features). Could anyone please share or point me to a clean code example for this? A working implementation would be a huge help for my research. Thank you for your time and help. Continue reading...
