Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Preprint |
| Published: |
2013
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/1301.1954 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866914848833536000 |
|---|---|
| author | Fishkind, Donniell E. Shen, Cencheng Park, Youngser Priebe, Carey E. |
| author_facet | Fishkind, Donniell E. Shen, Cencheng Park, Youngser Priebe, Carey E. |
| contents | Suppose that two large, multi-dimensional data sets are each noisy measurements of the same underlying random process, and principle components analysis is performed separately on the data sets to reduce their dimensionality. In some circumstances it may happen that the two lower-dimensional data sets have an inordinately large Procrustean fitting-error between them. The purpose of this manuscript is to quantify this "incommensurability phenomenon." In particular, under specified conditions, the square Procrustean fitting-error of the two normalized lower-dimensional data sets is (asymptotically) a convex combination (via a correlation parameter) of the Hausdorff distance between the projection subspaces and the maximum possible value of the square Procrustean fitting-error for normalized data. We show how this gives rise to the incommensurability phenomenon, and we employ illustrative simulations as well as a real data experiment to explore how the incommensurability phenomenon may have an appreciable impact. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_1301_1954 |
| institution | arXiv |
| publishDate | 2013 |
| record_format | arxiv |
| spellingShingle | On the Incommensurability Phenomenon Fishkind, Donniell E. Shen, Cencheng Park, Youngser Priebe, Carey E. Machine Learning Suppose that two large, multi-dimensional data sets are each noisy measurements of the same underlying random process, and principle components analysis is performed separately on the data sets to reduce their dimensionality. In some circumstances it may happen that the two lower-dimensional data sets have an inordinately large Procrustean fitting-error between them. The purpose of this manuscript is to quantify this "incommensurability phenomenon." In particular, under specified conditions, the square Procrustean fitting-error of the two normalized lower-dimensional data sets is (asymptotically) a convex combination (via a correlation parameter) of the Hausdorff distance between the projection subspaces and the maximum possible value of the square Procrustean fitting-error for normalized data. We show how this gives rise to the incommensurability phenomenon, and we employ illustrative simulations as well as a real data experiment to explore how the incommensurability phenomenon may have an appreciable impact. |
| title | On the Incommensurability Phenomenon |
| topic | Machine Learning |
| url | https://arxiv.org/abs/1301.1954 |