Research Projects

Quark and gluon jets from high-energy particle collisions produce distinct patterns in jet images, but distinguishing them remains a core challenge in particle physics. This project applies machine learning, especially GNNs, VAEs, and GVAEs, to detect patterns that differentiate quark- from gluon-origin jets.

Developing GNN-based models for predicting weather statistics using spatial-temporal graph representations. The project addresses challenges in streaming graph learning, including catastrophic forgetting and memory consumption optimization. We model weather stations as nodes with geographical connections to capture spatial dependencies in weather patterns.

Developing GNN-based approaches for simultaneous entity recognition and relation extraction from natural language text. The project aims to construct knowledge graphs by leveraging graph dependencies between words in sentences. Our model uses syntactic and semantic relationships to improve both entity identification and relation classification accuracy.

Applied Graph Neural Networks to predict various social network properties including average node degree, clustering coefficients, and average path lengths. The study analyzed how node and edge additions/deletions affect graph structure and compared GNN predictions with existing theoretical models and baseline approaches.

Developed a comprehensive software solution for online pharmaceutical database management. Implemented SRS documentation and prototype code for transaction management, including features for customer account management, digital transactions, credit payment systems, and transaction history tracking. Collaborated with team members as part of IACS Computer Science Unit project.