Media
AI for Health:
The file above is the proposal that was written on the topic of "Prototype Parts Based Classification to Improve Interpretability of Deep Neural Network Predictive Model". Below you will find the poster presentation that was given after the proposal:
Mapping the Research Frontier: Multi-granular Topic Modeling and Interactive Visualizations for Interdisciplinary Science:
This project develops a flexible, multi-granular framework to analyze and visualize interdisciplinary research by combining classical topic modeling with modern language embedding models and human-in-the-loop validation. We evaluate multiple modeling approaches to identify which best capture meaningful research themes across disciplines, and then translate the results into interactive visualizations, such as collaboration networks, topic evolution graphs, and cluster maps, that help researchers, program managers, and institutions explore scientific collaboration patterns, identify emerging areas of expertise, and discover potential interdisciplinary partnerships.
