High-Speed Function Approximation
| dc.contributor.author | Panda, Biswanath | en_US |
| dc.contributor.author | Riedewald, Mirek | en_US |
| dc.contributor.author | Gehrke, Johannes | en_US |
| dc.contributor.author | Pope, Stephen | en_US |
| dc.date.accessioned | 2007-09-04T15:45:56Z | |
| dc.date.available | 2007-09-04T15:45:56Z | |
| dc.date.issued | 2007-08-14 | en_US |
| dc.description.abstract | Learning methods for predictive models have traditionally focused on prediction quality and model building time, while prediction time(the time taken to make a prediction) is often ignored. However, there is an increasing need for models that are not only accurate, but also make fast predictions. Some of the most accurate models like ensemble models are often too slow to be used in practice. We believe that exploring the tradeoff between prediction time and model accuracy is an exciting new direction for data mining research. In this paper, we make a first step toward exploring this tradeoff. We introduce a new learning problem where we minimize model prediction time subject to a constraint on model accuracy. Our solution is a generic framework that leverages existing data mining algorithms while taking prediction time into account. We show a first application of our framework to a combustion simulation, and our results show significant improvements over existing methods. | en_US |
| dc.format.extent | 145888 bytes | |
| dc.format.mimetype | application/pdf | |
| dc.identifier.citation | http://techreports.library.cornell.edu:8081/Dienst/UI/1.0/Display/cul.cis/TR2007-2092 | en_US |
| dc.identifier.uri | https://hdl.handle.net/1813/8225 | |
| dc.language.iso | en_US | en_US |
| dc.publisher | Cornell University | en_US |
| dc.subject | computer science | en_US |
| dc.subject | Learning | en_US |
| dc.subject | technical report | en_US |
| dc.title | High-Speed Function Approximation | en_US |
| dc.type | technical report | en_US |
Files
Original bundle
1 - 1 of 1