Transfer Bounds for Linear Feature Learning
Abstract
If regression tasks are sampled from a distribution, then the expected error for a future task can be estimated by the average empirical errors on the data of a finite sample of tasks, uniformly over a class of regularizing or pre-processing transformations. The bound is dimension free, justifies optimization of the pre-processing feature-map and explains the circumstances under which learning-to-learn is preferable to single task learning.
Cite
Text
Maurer. "Transfer Bounds for Linear Feature Learning." Machine Learning, 2009. doi:10.1007/S10994-009-5109-7Markdown
[Maurer. "Transfer Bounds for Linear Feature Learning." Machine Learning, 2009.](https://mlanthology.org/mlj/2009/maurer2009mlj-transfer/) doi:10.1007/S10994-009-5109-7BibTeX
@article{maurer2009mlj-transfer,
title = {{Transfer Bounds for Linear Feature Learning}},
author = {Maurer, Andreas},
journal = {Machine Learning},
year = {2009},
pages = {327-350},
doi = {10.1007/S10994-009-5109-7},
volume = {75},
url = {https://mlanthology.org/mlj/2009/maurer2009mlj-transfer/}
}