Pairwise comparison is the process humans go through when they analyze the qualities of similar items to determine the best option. This works when there are only two or three objects to compare, but it becomes more complicated when dealing with massive data sets in social networks and other systems. CSL Assistant Professor Zhizhen Zhao is working on a method to compare a multitude of related items efficiently.
In many modern science and engineering fields, large-scale high dimensional data sets are generated with an abundance of structural information within each object. While this allows people to conduct detailed pairwise comparisons between individual objects, the data can be overwhelming.
In the recently funded project “Geometric Harmonic Analysis in Learning and Inference: Theory and Applications,” Zhao and team plan to analyze the manifold and data relationships to improve the efficacy and accuracy of computational data extraction using information already found within the data. This work could apply to data sets from social, biomedical, and comparative biological sciences.
“We want to organize (the items), and find communities among them,” said Zhao, assistant professor in electrical and computer engineering. “The measurements of the objects of interest are usually incomplete and noisy. We want to use harmonic analysis tools to efficiently extract structural information for the underlying objects through pairwise comparisons of the observation data. ”
The new models Zhao is developing could revolutionize how data is analyzed. That is why part of her project is to educate and train students on the models to better prepare them for their careers. In addition to educating students, Zhao and her team plan to use their models to disseminate research among various fields to build connections and organize workshops with a focus on data science efficacy.
This research is funded by NSF for three years in the amount of $210,000.