ICML 2015poster16 citations

Attribute Efficient Linear Regression with Distribution-Dependent Sampling

Doron Kukliansky, Ohad Shamir

Abstract

We consider a budgeted learning setting, where the learner can only choose and observe a small subset of the attributes of each training example. We develop efficient algorithms for Ridge and Lasso linear regression, which utilize the geometry of the data by a novel distribution-dependent sampling scheme, and have excess risk bounds which are better a factor of up to O(d/k) over the state-of-the-art, where d is the dimension and k+1 is the number of observed attributes per example. Moreover, under reasonable assumptions, our algorithms are the first in our setting which can provably use *less* attributes than full-information algorithms, which is the main concern in budgeted learning. We complement our theoretical analysis with experiments which support our claims.

BibTeX
@InProceedings{pmlr-v37-kukliansky15,
  title = 	 {Attribute Efficient Linear Regression with Distribution-Dependent Sampling},
  author = 	 {Kukliansky, Doron and Shamir, Ohad},
  booktitle = 	 {Proceedings of the 32nd International Conference on Machine Learning},
  pages = 	 {153--161},
  year = 	 {2015},
  editor = 	 {Bach, Francis and Blei, David},
  volume = 	 {37},
  series = 	 {Proceedings of Machine Learning Research},
  address = 	 {Lille, France},
  month = 	 {07--09 Jul},
  publisher =    {PMLR},
  pdf = 	 {http://proceedings.mlr.press/v37/kukliansky15.pdf},
  url = 	 {https://proceedings.mlr.press/v37/kukliansky15.html},
  abstract = 	 {We consider a budgeted learning setting, where the learner can only choose and observe a small subset of the attributes of each training example. We develop efficient algorithms for Ridge and Lasso linear regression, which utilize the geometry of the data by a novel distribution-dependent sampling scheme, and have excess risk bounds which are better a factor of up to O(d/k) over the state-of-the-art, where d is the dimension and k+1 is the number of observed attributes per example. Moreover, under reasonable assumptions, our algorithms are the first in our setting which can provably use *less* attributes than full-information algorithms, which is the main concern in budgeted learning. We complement our theoretical analysis with experiments which support our claims.}
}