Add dataset (use cicids2017) and train initial security model

[Abhi1366-create]

Task

Add a dataset for security-related analysis and train the initial machine learning model.

Description

The repository currently contains the basic project structure and a training script.
The next step is to add appropriate datasets and train the first model that can be used for security experimentation within the AI Security OS environment.

Steps

• Add a dataset to the datasets/ directory
• Prepare and preprocess the dataset if required
• Use the training script in training/ to train the model
• Save the trained model in the models/ directory
• Verify that the training script runs successfully

Expected Outcome

• Dataset available in the repository
• Successfully trained model
• Model stored in the models/ folder for future integration

[himelds]

Hi @Abhi1366-create, I’d like to work on this issue.

My plan is to:

• Add a suitable security-related dataset to the datasets/ directory
• Perform necessary preprocessing (handling missing values, encoding, feature selection)
• Use the existing training script to train a baseline classification model
• Save the trained model in the models/ directory
• Ensure the training pipeline runs end-to-end without errors

I will focus on creating a clean and reproducible baseline setup that can be extended later.

Please assign this issue to me if it’s available.

[Abhi1366-create]

@himelds himelds you are assigned use cicids data set

[himelds]

@Abhi1366-create Hi! I’ve completed the initial implementation for this issue.

• Added a cleaned subset of the CICIDS2017 dataset in the datasets/ directory
• Selected relevant numeric features and removed missing/infinite values
• Updated the training script to load data from the dataset instead of using hardcoded values
• Successfully trained an Isolation Forest model for anomaly detection

The training pipeline now runs end-to-end. The trained model is saved to the models/ directory during execution (not included in the repo to keep it lightweight).

Please let me know if any changes or improvements are needed. Thanks!

[Abhi1366-create]

Yeah looks fine !