Learning from Data with Heterogeneous Noise using SGD

Shuang Song, Kamalika Chaudhuri, Anand Sarwate
Proceedings of the Eighteenth International Conference on Artificial Intelligence and Statistics, PMLR 38:894-902, 2015.

Abstract

We consider learning from data of variable quality that may be obtained from different heterogeneous sources. Addressing learning from heterogeneous data in its full generality is a challenging problem. In this paper, we adopt instead a model in which data is observed through heterogeneous noise, where the noise level reflects the quality of the data source. We study how to use stochastic gradient algorithms to learn in this model. Our study is motivated by two concrete examples where this problem arises naturally: learning with local differential privacy based on data from multiple sources with different privacy requirements, and learning from data with labels of variable quality. The main contribution of this paper is to identify how heterogeneous noise impacts performance. We show that given two datasets with heterogeneous noise, the order in which to use them in standard SGD depends on the learning rate. We propose a method for changing the learning rate as a function of the heterogeneity, and prove new regret bounds for our method in two cases of interest. Finally, we evaluate the performance of our algorithm on real data.

Cite this Paper


BibTeX
@InProceedings{pmlr-v38-song15, title = {{Learning from Data with Heterogeneous Noise using SGD}}, author = {Shuang Song and Kamalika Chaudhuri and Anand Sarwate}, booktitle = {Proceedings of the Eighteenth International Conference on Artificial Intelligence and Statistics}, pages = {894--902}, year = {2015}, editor = {Guy Lebanon and S. V. N. Vishwanathan}, volume = {38}, series = {Proceedings of Machine Learning Research}, address = {San Diego, California, USA}, month = {09--12 May}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v38/song15.pdf}, url = { http://proceedings.mlr.press/v38/song15.html }, abstract = {We consider learning from data of variable quality that may be obtained from different heterogeneous sources. Addressing learning from heterogeneous data in its full generality is a challenging problem. In this paper, we adopt instead a model in which data is observed through heterogeneous noise, where the noise level reflects the quality of the data source. We study how to use stochastic gradient algorithms to learn in this model. Our study is motivated by two concrete examples where this problem arises naturally: learning with local differential privacy based on data from multiple sources with different privacy requirements, and learning from data with labels of variable quality. The main contribution of this paper is to identify how heterogeneous noise impacts performance. We show that given two datasets with heterogeneous noise, the order in which to use them in standard SGD depends on the learning rate. We propose a method for changing the learning rate as a function of the heterogeneity, and prove new regret bounds for our method in two cases of interest. Finally, we evaluate the performance of our algorithm on real data.} }
Endnote
%0 Conference Paper %T Learning from Data with Heterogeneous Noise using SGD %A Shuang Song %A Kamalika Chaudhuri %A Anand Sarwate %B Proceedings of the Eighteenth International Conference on Artificial Intelligence and Statistics %C Proceedings of Machine Learning Research %D 2015 %E Guy Lebanon %E S. V. N. Vishwanathan %F pmlr-v38-song15 %I PMLR %P 894--902 %U http://proceedings.mlr.press/v38/song15.html %V 38 %X We consider learning from data of variable quality that may be obtained from different heterogeneous sources. Addressing learning from heterogeneous data in its full generality is a challenging problem. In this paper, we adopt instead a model in which data is observed through heterogeneous noise, where the noise level reflects the quality of the data source. We study how to use stochastic gradient algorithms to learn in this model. Our study is motivated by two concrete examples where this problem arises naturally: learning with local differential privacy based on data from multiple sources with different privacy requirements, and learning from data with labels of variable quality. The main contribution of this paper is to identify how heterogeneous noise impacts performance. We show that given two datasets with heterogeneous noise, the order in which to use them in standard SGD depends on the learning rate. We propose a method for changing the learning rate as a function of the heterogeneity, and prove new regret bounds for our method in two cases of interest. Finally, we evaluate the performance of our algorithm on real data.
RIS
TY - CPAPER TI - Learning from Data with Heterogeneous Noise using SGD AU - Shuang Song AU - Kamalika Chaudhuri AU - Anand Sarwate BT - Proceedings of the Eighteenth International Conference on Artificial Intelligence and Statistics DA - 2015/02/21 ED - Guy Lebanon ED - S. V. N. Vishwanathan ID - pmlr-v38-song15 PB - PMLR DP - Proceedings of Machine Learning Research VL - 38 SP - 894 EP - 902 L1 - http://proceedings.mlr.press/v38/song15.pdf UR - http://proceedings.mlr.press/v38/song15.html AB - We consider learning from data of variable quality that may be obtained from different heterogeneous sources. Addressing learning from heterogeneous data in its full generality is a challenging problem. In this paper, we adopt instead a model in which data is observed through heterogeneous noise, where the noise level reflects the quality of the data source. We study how to use stochastic gradient algorithms to learn in this model. Our study is motivated by two concrete examples where this problem arises naturally: learning with local differential privacy based on data from multiple sources with different privacy requirements, and learning from data with labels of variable quality. The main contribution of this paper is to identify how heterogeneous noise impacts performance. We show that given two datasets with heterogeneous noise, the order in which to use them in standard SGD depends on the learning rate. We propose a method for changing the learning rate as a function of the heterogeneity, and prove new regret bounds for our method in two cases of interest. Finally, we evaluate the performance of our algorithm on real data. ER -
APA
Song, S., Chaudhuri, K. & Sarwate, A.. (2015). Learning from Data with Heterogeneous Noise using SGD. Proceedings of the Eighteenth International Conference on Artificial Intelligence and Statistics, in Proceedings of Machine Learning Research 38:894-902 Available from http://proceedings.mlr.press/v38/song15.html .

Related Material