Learning Classes Correlated to a Hierarchy
Trees are a common way of organizing large amounts of information by placing items with similar characteristics near one another in the tree. We introduce a classification problem where a given tree structure gives us information on the best way to label nearby elements. We suggest there are ma...
| Main Authors: | , |
|---|---|
| Language: | en_US |
| Published: |
2004
|
| Online Access: | http://hdl.handle.net/1721.1/6719 |
| _version_ | 1826201794059960320 |
|---|---|
| author | Shih, Lawrence Karger, David |
| author_facet | Shih, Lawrence Karger, David |
| author_sort | Shih, Lawrence |
| collection | MIT |
| description | Trees are a common way of organizing large amounts of information by placing items with similar characteristics near one another in the tree. We introduce a classification problem where a given tree structure gives us information on the best way to label nearby elements. We suggest there are many practical problems that fall under this domain. We propose a way to map the classification problem onto a standard Bayesian inference problem. We also give a fast, specialized inference algorithm that incrementally updates relevant probabilities. We apply this algorithm to web-classification problems and show that our algorithm empirically works well. |
| first_indexed | 2024-09-23T11:57:00Z |
| id | mit-1721.1/6719 |
| institution | Massachusetts Institute of Technology |
| language | en_US |
| last_indexed | 2024-09-23T11:57:00Z |
| publishDate | 2004 |
| record_format | dspace |
| spelling | mit-1721.1/67192019-04-11T02:53:04Z Learning Classes Correlated to a Hierarchy Shih, Lawrence Karger, David Trees are a common way of organizing large amounts of information by placing items with similar characteristics near one another in the tree. We introduce a classification problem where a given tree structure gives us information on the best way to label nearby elements. We suggest there are many practical problems that fall under this domain. We propose a way to map the classification problem onto a standard Bayesian inference problem. We also give a fast, specialized inference algorithm that incrementally updates relevant probabilities. We apply this algorithm to web-classification problems and show that our algorithm empirically works well. 2004-10-08T20:38:58Z 2004-10-08T20:38:58Z 2003-05-01 AIM-2003-013 http://hdl.handle.net/1721.1/6719 en_US AIM-2003-013 1146195 bytes 480357 bytes application/postscript application/pdf application/postscript application/pdf |
| spellingShingle | Shih, Lawrence Karger, David Learning Classes Correlated to a Hierarchy |
| title | Learning Classes Correlated to a Hierarchy |
| title_full | Learning Classes Correlated to a Hierarchy |
| title_fullStr | Learning Classes Correlated to a Hierarchy |
| title_full_unstemmed | Learning Classes Correlated to a Hierarchy |
| title_short | Learning Classes Correlated to a Hierarchy |
| title_sort | learning classes correlated to a hierarchy |
| url | http://hdl.handle.net/1721.1/6719 |
| work_keys_str_mv | AT shihlawrence learningclassescorrelatedtoahierarchy AT kargerdavid learningclassescorrelatedtoahierarchy |