TensorLEGO: Contextualized Tensor Analysis for Complex Data
Abstract
We will develop a general tensor analysis framework for mining big data with complex relationships. Specifically, rather than analyzing data tensor alone, we perform contextualization of tensor analysis to take into consideration the complex relationships among the data points. We propose four concrete frameworks to handle different types of contexts. They are homogeneous contextualization, heterogeneous contextualization, temporal contextualization and distributed contextualization. We also leveraged the tensor network elements in quantum physics to graphically interpret the different contextualizations, so that the complicated tensor decompositions can be exploited in a plug-and-play framework like LEGO. We will evaluate the proposed methodologies in different medical applications because of the complex nature of health problems. We believe such framework will be very helpful and valuable to military applications involving complex big data analysis.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Jul 26, 2018
- Source ID
- N000141812585
Entities
People
- Fei Wang
Organizations
- Office of Naval Research
- United States Navy
- Weill Cornell Medicine