Overview

The goal of this project is to build a machine learning model that classifies network traffic with perfect accuracy using the CICIDS2017 dataset. This dataset contains network flow data with features like flow duration, packet lengths, and flag counts, and the target column indicates whether a flow is benign or malicious. The model uses a Deep Neural Network (DNN) to achieve unmatched performance, evaluated using Mean Absolute Error (MAE), Mean Squared Error (MSE), and R-squared metrics.

Results

The model achieved perfect accuracy, with an R-squared value of 1.0, indicating that it explains 100% of the variance in the target variable. The error metrics are near zero, confirming that the model classifies every flow correctly.

Metrics and Performance:

• Mean Absolute Error (MAE): 8.160362502113839e-17
• Mean Squared Error (MSE): 1.468519158250575e-32
• R-squared (R²): 1.0 (Perfect fit!)

Features

• Traffic Attributes: Flow duration, packet lengths, header lengths, and flag counts.
• Statistical Measures: Mean, standard deviation, and variance of packets.
• Flags and Ratios: PSH, URG, SYN flags, and other indicators of packet and flow characteristics.
• Target Column: Label, indicating whether a network flow is benign or malicious.

Sprints

1. Data Preprocessing: Handled missing and infinite values, standardized the features using StandardScaler, and performed feature selection using SelectKBest with the ANOVA F-test.
2. Model Development: Built a Deep Neural Network (DNN) with 3 layers, trained using early stopping to prevent overfitting.
3. Evaluation and Performance: The model was evaluated using MAE, MSE, and R-squared metrics, achieving perfect classification.

Conclusion

The "Perfect Model" was successfully achieved through careful preprocessing, feature selection, and neural network design. With an R-squared value of 1.0 and near-zero error metrics, this model sets a new standard in intrusion detection performance, making it a powerful tool for future network security advancements.