The security and reliability of smart contracts are critical issues in the blockchain ecosystem. Existing methods often employ machine learning techniques and static analysis, which have limitations, such as computational inefficiency and incomplete code analysis.
We introduce VdaBSc, a comprehensive approach incorporating dynamic analysis, real-time runtime batch normalization, data augmentation, N-grams, and a hybrid architecture combining BiLSTM, CNN, and the Attention Mechanism to address these challenges.
Our feature representation technique employs N-grams and one-hot encoding, capturing sequential dependencies between opcodes and representing each opcode as a binary vector.
The VdaBSc model is built on a robust hybrid architecture that leverages BiLSTM for capturing temporal dynamics, CNN for local feature extraction, and the Attention Mechanism for context understanding.
VdaBSc has been rigorously evaluated against existing methods and state-of-the-art deep learning techniques. Our evaluations, benchmarked against these methods, reveal that VdaBSc demonstrates superior performance across key metrics such as accuracy, precision, recall, and F1-Score.
Before you begin, ensure you have the following installed:
- Python (version 3.7 or higher)
- Git (for cloning the repository)
-
Clone the Repository
git clone https://github.com/niirex1/VdaBSc-project.git
-
Navigate to the Project Directory
cd VdaBSc-project -
Install Required Python Packages
pip install -r requirements.txt
Please refer to the example_usage.ipynb notebook for a detailed guide on how to use the model for smart contract vulnerability detection.
We welcome contributions from the research community. For guidelines on contributing, please refer to the Contributing documentation.
VdaBSc is licensed under the MIT License. For the full license text, refer to the LICENSE file in the repository or visit MIT License.
For any questions, feedback, or suggestions regarding the VdaBSc project, please reach out to the project maintainers:
- Rexford Sosu
- Email: rexfordsosu@outlook.com
- GitHub: @rexfordsosu
- LinkedIn: Rexford's LinkedIn
We appreciate your interest in the VdaBSc project and look forward to your contributions!