Methods of Sparse Modeling and Dimensionality Reduction to Deal with Big Data
Abstract
This project focused on the development of new methods for sparse modeling and non-negative matrix factorization. The major achievements are 1) a sparse topic model that can learn thousands of topics from a large set of documents and infer the topic mixture of each document, 2) a supervised dimension reduction method for large datasets, and 3) a non-negative matrix factorization (NMF) method with good interpretability. The research on sparse topic model and supervised dimension reduction as well as NMF is motivated by the need of reducing complexity in dealing with huge and complex datasets in big data. The proposed methods were theoretically and experimentally evaluated, and applied to problems in materials science and biomedicine.
Document Details
- Document Type
- Technical Report
- Publication Date
- Apr 01, 2015
- Accession Number
- ADA623178
Entities
People
- Tu Bao Ho
Organizations
- Japan Advanced Institute of Science and Technology