Given several fragments of a dataset (distributed across several sites), a set of constraints (referring to privacy concerns, storage issues, operations allowed, etc.), a hypothesis class, and a performance criterion, the task of the distributed learner is to output a hypothesis that optimizes the performance criterion without violating the set of constraints. We say that an algorithm for learning from distributed data is exact if the hypothesis that it outputs is identical to that produced by the centralized algorithm for learning from the complete dataset, obtained by appropriately combining the distributed data fragments.
Published in Chapter:
Learning Classifiers from Distributed Data Sources
Doina Caragea (Kansas State University, USA) and Vasant Honavar (Iowa State University, USA)
Copyright: © 2009
|Pages: 8
DOI: 10.4018/978-1-60566-242-8.ch063
Abstract
Recent development of high throughput data acquisition technologies in a number of domains (e.g., biological sciences, atmospheric sciences, space sciences, commerce) together with advances in digital storage, computing, and communications technologies have resulted in the proliferation of a multitude of physically distributed data repositories created and maintained by autonomous entities (e.g., scientists, organizations). The resulting increasingly data-rich domains offer unprecedented opportunities in computer assisted data-driven knowledge acquisition in a number of applications, including, in particular, data-driven scientific discovery, data-driven decision-making in business and commerce, monitoring and control of complex systems, and security informatics.