Saved in:
| Main Author: | |
|---|---|
| Format: | Recurso digital |
| Language: | |
| Published: |
Zenodo
2025
|
| Online Access: | https://doi.org/10.5281/zenodo.18021224 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866901677832929280 |
|---|---|
| author | Aoudi, Samer |
| author_facet | Aoudi, Samer |
| contents | <h2> Release Notes</h2> <p>This patch release finalizes the repository for integration with <strong>Zenodo</strong>, generating a permanent Digital Object Identifier (DOI) for academic citation.</p> <h3> What's New in v1.0.1</h3> <ul> <li><strong>Zenodo Integration:</strong> Configured repository metadata for permanent archiving.</li> <li><strong>Documentation:</strong> Added status badges (DOI, License) to the <code>README.md</code>.</li> <li><strong>Stability:</strong> Codebase is frozen to match the manuscript submission state.</li> </ul> <h2> Framework Overview</h2> <p>The <strong>Adversarial Robustness Evaluation of AI IDS Framework</strong> provides a reproducible pipeline to evaluate the robustness of Machine Learning (Random Forest, Logistic Regression) and Deep Learning (MLP, CNN-1D) Intrusion Detection Systems against adversarial attacks under realistic black-box transfer conditions.</p> <h2>Key Features</h2> <ul> <li><strong>End-to-End Preprocessing:</strong> Automated cleaning, normalization, and stratified splitting (70/10/20) for <strong>CICIDS2017</strong> and <strong>CICIDS2018</strong> datasets.</li> <li><strong>Model Training:</strong> Scripts to train and evaluate:<ul> <li><strong>Baselines:</strong> Random Forest (RF), Logistic Regression (LR).</li> <li><strong>Deep Learning:</strong> Multilayer Perceptron (MLP), 1D-CNN.</li> <li><strong>Surrogate Models:</strong> Independent shadow models for black-box attack generation.</li> </ul> </li> <li><strong>Adversarial Attack Suite:</strong> Implementation of four attack families with <strong>Semantic Constraints</strong> (feature validity enforcement):<ul> <li><strong>Gradient-Based:</strong> FGSM, PGD ($L_{\infty}$ norm).</li> <li><strong>Black-Box:</strong> HopSkipJump (HSJA), Zeroth-Order Optimization (ZOO).</li> </ul> </li> <li><strong>Cross-Dataset Evaluation:</strong> Tools to test the transferability of adversarial examples from CICIDS2017 to the CICIDS2018 Friday slice.</li> </ul> <h2>Quick Start</h2> <ol> <li><strong>Clone the repository:</strong><pre><code>git clone [https://github.com/YourUsername/IDS-Adversarial-Robustness.git](https://github.com/sameroudi/adversarial-evaluation-ai-ids.git) </code></pre> </li> <li><strong>Install dependencies:</strong><pre><code>pip install -r requirements.txt </code></pre> </li> <li><strong>Run the pipeline:</strong><pre><code># 1. Preprocess Data python scripts/prepare_cicids2017.py # 2. Train Models python scripts/train_deep_cicids2017.py # 3. Execute Attacks python scripts/run_attacks_cicids2017.py </code></pre> </li> </ol> <h2>Citation</h2> <p>Please cite the software using the generated DOI:</p> <blockquote> <p>Aoudi, S. (2025). <em>Adversarial Robustness Evaluation of AI IDS Framework (v1.0.1)</em> [Computer software]. Zenodo. https://doi.org/10.5281/zenodo.17999876</p> </blockquote> |
| format | Recurso digital |
| id | zenodo_https___doi_org_10_5281_zenodo_18021224 |
| institution | Zenodo |
| language | |
| publishDate | 2025 |
| publisher | Zenodo |
| record_format | zenodo |
| spellingShingle | Adversarial Robustness of AI Intrusion Detection Systems Aoudi, Samer <h2> Release Notes</h2> <p>This patch release finalizes the repository for integration with <strong>Zenodo</strong>, generating a permanent Digital Object Identifier (DOI) for academic citation.</p> <h3> What's New in v1.0.1</h3> <ul> <li><strong>Zenodo Integration:</strong> Configured repository metadata for permanent archiving.</li> <li><strong>Documentation:</strong> Added status badges (DOI, License) to the <code>README.md</code>.</li> <li><strong>Stability:</strong> Codebase is frozen to match the manuscript submission state.</li> </ul> <h2> Framework Overview</h2> <p>The <strong>Adversarial Robustness Evaluation of AI IDS Framework</strong> provides a reproducible pipeline to evaluate the robustness of Machine Learning (Random Forest, Logistic Regression) and Deep Learning (MLP, CNN-1D) Intrusion Detection Systems against adversarial attacks under realistic black-box transfer conditions.</p> <h2>Key Features</h2> <ul> <li><strong>End-to-End Preprocessing:</strong> Automated cleaning, normalization, and stratified splitting (70/10/20) for <strong>CICIDS2017</strong> and <strong>CICIDS2018</strong> datasets.</li> <li><strong>Model Training:</strong> Scripts to train and evaluate:<ul> <li><strong>Baselines:</strong> Random Forest (RF), Logistic Regression (LR).</li> <li><strong>Deep Learning:</strong> Multilayer Perceptron (MLP), 1D-CNN.</li> <li><strong>Surrogate Models:</strong> Independent shadow models for black-box attack generation.</li> </ul> </li> <li><strong>Adversarial Attack Suite:</strong> Implementation of four attack families with <strong>Semantic Constraints</strong> (feature validity enforcement):<ul> <li><strong>Gradient-Based:</strong> FGSM, PGD ($L_{\infty}$ norm).</li> <li><strong>Black-Box:</strong> HopSkipJump (HSJA), Zeroth-Order Optimization (ZOO).</li> </ul> </li> <li><strong>Cross-Dataset Evaluation:</strong> Tools to test the transferability of adversarial examples from CICIDS2017 to the CICIDS2018 Friday slice.</li> </ul> <h2>Quick Start</h2> <ol> <li><strong>Clone the repository:</strong><pre><code>git clone [https://github.com/YourUsername/IDS-Adversarial-Robustness.git](https://github.com/sameroudi/adversarial-evaluation-ai-ids.git) </code></pre> </li> <li><strong>Install dependencies:</strong><pre><code>pip install -r requirements.txt </code></pre> </li> <li><strong>Run the pipeline:</strong><pre><code># 1. Preprocess Data python scripts/prepare_cicids2017.py # 2. Train Models python scripts/train_deep_cicids2017.py # 3. Execute Attacks python scripts/run_attacks_cicids2017.py </code></pre> </li> </ol> <h2>Citation</h2> <p>Please cite the software using the generated DOI:</p> <blockquote> <p>Aoudi, S. (2025). <em>Adversarial Robustness Evaluation of AI IDS Framework (v1.0.1)</em> [Computer software]. Zenodo. https://doi.org/10.5281/zenodo.17999876</p> </blockquote> |
| title | Adversarial Robustness of AI Intrusion Detection Systems |
| url | https://doi.org/10.5281/zenodo.18021224 |