Measuring K-Wise Independence for Large Data Sets
|
 |
By Bob Nidever in Latest Inventions Published: Friday, 14 November 08 - 10:12 PM (GMT) |
| ||
UCLA scientists have developed the first approach to measure correlations in large volumes of data streams quickly and efficiently. Data streams are ubiquitous in marketing, financial, security, sensor, web, biological, and communication applications, and the novel algorithm described here a major breakthrough in data mining and prediction.
Researchers at UCLA have created the first algorithm for measuring correlations in continuous and large data streams. This algorithm represents a novel combinatorial approach to analyzing second moment, or variance of dependent sketches of data streams. Correlations between multidimensional data can readily be acquired, while maintaining efficient dimensionality reduction. MORE [2008-759]
<-- Back
