 |
Center for Nonlinear Studies |
High-throughput imaging instruments that generate more than 1TB of data per day have created a “Big Data” crisis in neuroscience in which the size and complexity of data exceed the capability of labs to manage it. The Open Connectome Project defines a community platform that manages massive brain-imaging data sets in exchange for providing open access. The Open Connectome Project stores data in high-dimensional spatial databases on data-intensive clusters and provides Web services that link the data to high-performance computing. On this architecture, we run parallel computer vision algorithms that extract, store, and analyze brain structure. As a platform for Open Science, the Open Connectome Project democratizes access to world-class brain data, making it publicly available to an interdisciplinary community of researchers, including statisticians, physicists, and computer scientists. This talk will describe how data-intensive computing is transforming Open (Neuro)-Science. It will cover the hardware and software architecture of the services, including spatial queries, data representations and placement, and integration with parallel computing.
Host: Josephine Olivas