For many types of learners one can compute the statistically 'optimal' way to select data. We review how these techniques have been used with feedforward neural networks. We then show how the same principles may be used to select data for two alternative, statistically-based learning architectures: mixtures of Gaussians and locally weighted regression. While the techniques for neural networks are expensive and approximate, the techniques for mixtures of Gaussians and locally weighted regression are both efficient and accurate.
| Identifer | oai:union.ndltd.org:MIT/oai:dspace.mit.edu:1721.1/7192 |
| Date | 21 March 1995 |
| Creators | Cohn, David A., Ghahramani, Zoubin, Jordan, Michael I. |
| Source Sets | M.I.T. Theses and Dissertation |
| Language | en_US |
| Detected Language | English |
| Format | 6 p., 266098 bytes, 440905 bytes, application/postscript, application/pdf |
| Relation | AIM-1522, CBCL-110 |
Page generated in 0.0018 seconds