Researching in the field of Abnormal Traffic Identification Methods.
Hello Tarnue, can you please give us more information about your research project?
1. What is the problem you want to solve?
2. How do other researchers solve it and how do you want to differentiate yourself?
3. Do you have data for this topic?
Thanks Roman,
Basically I would like to look at three major problems which most research are using supervised Machine learning methods to identify these abnormal traffic;
1. feature identification and recognition.
2. Gain higher accuracy.
3. Find out algorithm reaction to new indirect attacks.
A new and innovative way to gain insight into sensitive features and models is interpretable machine learning. The shap library is highly recommended for this purpose: https://github.com/slundberg/shap
Your procedure could look like this:
1. Collect features from literature
2. Add your features
3. Create for example a xgboost model and check if the relationship can be modelled well
4. If yes -> perform the shap analysis
You will learn how sensitive individual features are and how the interactions look like.
- Badges
- Science topic