Public Release: 

Building a stronger 'brain' for brain data research

Pitt computer engineer receives $1.2 million NSF award to make big brain data collaborations more secure

University of Pittsburgh

IMAGE

IMAGE: Heng Huang, John A. Jurenko Professor of Computer Engineering at the University of Pittsburgh Swanson School of Engineering. view more 

Credit: Swanson School of Engineering

PITTSBURGH (December 20, 2018) ... The National Science Foundation BIGDATA program awarded $1,200,000 to a research team led by the University of Pittsburgh Swanson School of Engineering to study the big brain data for complex brain disorders and design new algorithms that address computational challenges in multi-site collaborative data mining.

Heng Huang, the Swanson School's John A. Jurenko Professor of Computer Engineering, is principal investigator of the study, "Asynchronous Distributed Machine Learning Framework for Multi-Site Collaborative Brain Big Data Mining." Huang currently leads seven NSF projects and an NIH R01 project on machine learning, big data mining, computational neuroscience, health informatics, and precision medicine.

"Research in emerging fields, such as brain imaging genomics and human connectomics, holds great promise for a systems biology study of the brain," said Huang. "This research can help us better understand complex neurobiological systems, from genetic determinants to the interplay between brain structure, connectivity, function, and cognition."

While researchers currently have access to brain data collected from a series of funded projects, they have failed to attain additional data collected by different local institutes due to data privacy and security issues preventing cross-institutional distribution. In this project, Huang will create a framework to address these issues and facilitate data and computing resource sharing.

"In collaborative data analysis, the participating institutes keep their own data, which are analyzed and computed locally, and only share the computed results via communicating with the server," explained Huang. "The server communicates with all institutes and updates the computational model such that the trained machine learning models indirectly use all data and are shared to all institutes."

According to Huang, most machine learning algorithms were not designed for such distributed architecture due to difficulties in designing efficient algorithms and providing theoretical foundations. This is the first project to create these type of algorithms for the study of brain imaging genomics and human connectomics. The goal of this project is to alleviate these computational challenges and enable investigators in neuroimaging, genomics, neuroscience, and other brain-related disciplines to securely and more efficiently further their research.

"The result of our project will be new distributed machine learning algorithms with theoretical foundations that can be used for multi-site collaborative big brain data mining, creating large-scale computational strategies and effective software tools," said Huang. "We hope that this work will help researchers harness the full potential of big brain data, potentially leading to the next major brain science discoveries."

###

Disclaimer: AAAS and EurekAlert! are not responsible for the accuracy of news releases posted to EurekAlert! by contributing institutions or for the use of any information through the EurekAlert system.