Development of a Firewall for the POX SDN Controller
Designed and implemented a Python-based firewall module tailored for the POX SDN Controller, with comprehensive testing conducted within the Mininet simulation environment.
Automata Learning using the LearnLib Framework
Experienced in automata learning using the LearnLib framework, specifically employing the L* and TTT algorithms for active learning and system verification.
COMPAS Recidivism Racial Bias and Model Explainability with LIME and SHAP
Demonstrated that the algorithm exhibits bias in favor of white defendants and against black inmates.
Reimplementation of the paper ‘Synthesis and Machine Learning for Heterogeneous Extraction’
Combined techniques from the Software Synthesis and Machine Learning to extract structured information from heterogeneous data.
Extension and Further Analysis of Contrastive Framework in the Task of Text Summarization
Investigated the performance of the contrastive framework in the task of summarization, making the representation space of the language model more isotropic, which was then leveraged to generate more diverse texts.
Tested Java Source Code using the Randoop and EvoSuite
Evaluated Java source code using the Randoop and EvoSuite tools, harnessing their automated testing capabilities to ensure code robustness and functionality.
Microsoft Malware Prediction
Predicted if a machine will soon be hit with malware or not using Machine Learning.
New York City Taxi Trip Duration Prediction
Built a model that predicts the total ride duration of taxi trips in New York City.
Real-time Augmented Reality
The transformation is derived from the homography between the reference surface coordinate system and the target image coordinate system, allowing for the projection of the 3D model into the image’s pixel space.