Abstract:
Security applications employ machine learning models. However, attackers can adapt to avoid detection. Machine learning models have been strengthened against such attacks. However, randomization is underutilized. Most studies use models with differentiable error functions, but tree-based models are popular because they are easy to interpret. This paper proposes a novel randomization-based approach to make tree-based models more evasion-resistant. The proposed method uses randomization during model training and model application (detecting attacks). This method is also applied to random forest, an ML method that incorporates randomness at training time but often fails to produce robust models. We developed a weighted-random-forest method to create more robust models and a clustering method to add randomness at model application time. We proposed a theoretical framework to lower adversaries’ effort. Our approach improves random-forest robustness in intrusion detection and spam filtering experiments.
Note: Please discuss with our team before submitting this abstract to the college. This Abstract or Synopsis varies based on student project requirements.
Did you like this final year project?
To download this project Code with thesis report and project training... Click Here
