Robustness Metrics for Tumor Volume Delineation CNN
Designed robustness metrics for the nnU-Net medical imaging model, using MONAI to test performance against severe data augmentations on CT scans.
Selected Projects
Beyond my primary research in numerical analysis, I actively explore applications in machine learning and deep learning. Below are selected projects demonstrating my work with neural networks, natural language processing, and data mining.
Multimodal Dialogue Generation for The Office TV Show
A unified Deep Learning pipeline combining CNNs (VGG16), CLIP, and GPT-2 to generate character-specific dialogue based on scene context.
Optimal Coefficients of Runge-Kutta Schemes with Machine Learning
Used Neural Networks (TensorFlow/Keras) to discover optimal coefficients for Runge-Kutta ODE solvers, reproducing classical schemes and improving efficiency for stiff problems.
Detecting Credit Card Fraud with Machine Learning
Collaborated with large multinational financial institution through PIC Math to implement Random Forest and KNN algorithms for fraud detection on large-scale datasets.