CODEWITHSUNDEEP
normalizationautoencoderanomaly-detectionintrusion-detectionids
Vincent Oscar
Vincent Oscar

Reputation: 1

Best normalization method for auto-encoder

I am building an intrusion detection system to detect malicious traffic in my network using an AutoEncoder. I've been training my model to learn from benign traffic and minimize the Mean Squared Error (MSE) as much as possible. At the same time, I've aimed to prevent overfitting by ensuring the model doesn't simply memorize the input data. When I saved my model and tested it with different data (such as an ICMP flood attack), it yielded an error rate lower than that for benign traffic. In benign traffic, the ICMP type is either 0 (reply) or 8 (request), but in ICMP flood data, it's solely type 8.

I tried using normalization methods such as MinMaxScaler and StandardScaler, but they didn't seem to make a significant difference. I'm looking for a more effective normalization method to improve the performance of my AutoEncoder results.

Upvotes: 0

Views: 39

Answers (0)

Related Questions