Selective sampling algorithms for cost-sensitive multiclass prediction

Alekh Agarwal
; Proceedings of the 30th International Conference on Machine Learning, PMLR 28(3):1220-1228, 2013.

Abstract

In this paper, we study the problem of active learning for cost-sensitive multiclass classification. We propose selective sampling algorithms, which process the data in a streaming fashion, querying only a subset of the labels. For these algorithms, we analyze the regret and label complexity when the labels are generated according to a generalized linear model. We establish that the gains of active learning over passive learning can range from none to exponentially large, based on a natural notion of margin. We also present a safety guarantee to guard against model mismatch. Numerical simulations show that our algorithms indeed obtain a low regret with a small number of queries.

Cite this Paper


BibTeX
@InProceedings{pmlr-v28-agarwal13, title = {Selective sampling algorithms for cost-sensitive multiclass prediction}, author = {Alekh Agarwal}, booktitle = {Proceedings of the 30th International Conference on Machine Learning}, pages = {1220--1228}, year = {2013}, editor = {Sanjoy Dasgupta and David McAllester}, volume = {28}, number = {3}, series = {Proceedings of Machine Learning Research}, address = {Atlanta, Georgia, USA}, month = {17--19 Jun}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v28/agarwal13.pdf}, url = {http://proceedings.mlr.press/v28/agarwal13.html}, abstract = {In this paper, we study the problem of active learning for cost-sensitive multiclass classification. We propose selective sampling algorithms, which process the data in a streaming fashion, querying only a subset of the labels. For these algorithms, we analyze the regret and label complexity when the labels are generated according to a generalized linear model. We establish that the gains of active learning over passive learning can range from none to exponentially large, based on a natural notion of margin. We also present a safety guarantee to guard against model mismatch. Numerical simulations show that our algorithms indeed obtain a low regret with a small number of queries.} }
Endnote
%0 Conference Paper %T Selective sampling algorithms for cost-sensitive multiclass prediction %A Alekh Agarwal %B Proceedings of the 30th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2013 %E Sanjoy Dasgupta %E David McAllester %F pmlr-v28-agarwal13 %I PMLR %J Proceedings of Machine Learning Research %P 1220--1228 %U http://proceedings.mlr.press %V 28 %N 3 %W PMLR %X In this paper, we study the problem of active learning for cost-sensitive multiclass classification. We propose selective sampling algorithms, which process the data in a streaming fashion, querying only a subset of the labels. For these algorithms, we analyze the regret and label complexity when the labels are generated according to a generalized linear model. We establish that the gains of active learning over passive learning can range from none to exponentially large, based on a natural notion of margin. We also present a safety guarantee to guard against model mismatch. Numerical simulations show that our algorithms indeed obtain a low regret with a small number of queries.
RIS
TY - CPAPER TI - Selective sampling algorithms for cost-sensitive multiclass prediction AU - Alekh Agarwal BT - Proceedings of the 30th International Conference on Machine Learning PY - 2013/02/13 DA - 2013/02/13 ED - Sanjoy Dasgupta ED - David McAllester ID - pmlr-v28-agarwal13 PB - PMLR SP - 1220 DP - PMLR EP - 1228 L1 - http://proceedings.mlr.press/v28/agarwal13.pdf UR - http://proceedings.mlr.press/v28/agarwal13.html AB - In this paper, we study the problem of active learning for cost-sensitive multiclass classification. We propose selective sampling algorithms, which process the data in a streaming fashion, querying only a subset of the labels. For these algorithms, we analyze the regret and label complexity when the labels are generated according to a generalized linear model. We establish that the gains of active learning over passive learning can range from none to exponentially large, based on a natural notion of margin. We also present a safety guarantee to guard against model mismatch. Numerical simulations show that our algorithms indeed obtain a low regret with a small number of queries. ER -
APA
Agarwal, A.. (2013). Selective sampling algorithms for cost-sensitive multiclass prediction. Proceedings of the 30th International Conference on Machine Learning, in PMLR 28(3):1220-1228

Related Material