Scientific data analysis and visualization via topology guided feature extraction and comparison

Efficient analysis and interactive visual exploration techniques are necessary for modern scientific research, which heavily relies on large and complex datasets generated from simulations or experiments.

This project aims to investigate feature-based analysis methods for scientific data that are built on the robust mathematical foundations of topological data analysis techniques.

The initial step involves feature extraction, which entails identifying and extracting hidden substructures in the data.

To accomplish this, we will develop novel multi-scale feature descriptors based on the powerful data abstraction capabilities of topology.

However, the primary focus of our project will be on the subsequent step of feature comparison which we argue holds the key to a paradigm shift in scientific data analysis and visualization as it enables a variety of downstream analysis tasks such as data summarization, feature tracking, and ensemble analysis.

We will create and implement innovative comparison measures that are both robust and computationally efficient.

Our proposed work will make a significant contribution to the current state-of-the-art in topological data analysis and scientific visualization, while simultaneously it will impact a wide range of scientific disciplines by providing new insights and knowledge through the analysis of feature-rich datasets.