gLOP: the global and Local Penalty for Capturing Predictive Heterogeneity

Rhiannon Rose, Daniel Lizotte
Proceedings of the 1st Machine Learning for Healthcare Conference, PMLR 56:134-149, 2016.

Abstract

When faced with a supervised learning problem, we hope to have rich enough data to build a model that predicts future instances well. However, in practice, problems can exhibit predictive heterogeneity: most instances might be relatively easy to predict, while others might be predictive outliers for which a model trained on the entire dataset does not perform well. Identifying these can help focus future data collection. We present gLOP, the global and Local Penalty, a framework for capturing predictive heterogeneity and identifying predictive outliers. gLOP is based on penalized regression for multitask learning, which improves learning by leveraging training signal information from related tasks. We give two optimization algorithms for gLOP, one space-efficient, and another giving the full regularization path. We also characterize uniqueness in terms of the data and tuning parameters, and present empirical results on synthetic data and on two health research problems.

Cite this Paper

BibTeX
@InProceedings{pmlr-v56-Rose16, title = {gLOP: the global and Local Penalty for Capturing Predictive Heterogeneity}, author = {Rose, Rhiannon and Lizotte, Daniel}, booktitle = {Proceedings of the 1st Machine Learning for Healthcare Conference}, pages = {134--149}, year = {2016}, editor = {Doshi-Velez, Finale and Fackler, Jim and Kale, David and Wallace, Byron and Wiens, Jenna}, volume = {56}, series = {Proceedings of Machine Learning Research}, address = {Northeastern University, Boston, MA, USA}, month = {18--19 Aug}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v56/Rose16.pdf}, url = {https://proceedings.mlr.press/v56/Rose16.html}, abstract = {When faced with a supervised learning problem, we hope to have rich enough data to build a model that predicts future instances well. However, in practice, problems can exhibit predictive heterogeneity: most instances might be relatively easy to predict, while others might be predictive outliers for which a model trained on the entire dataset does not perform well. Identifying these can help focus future data collection. We present gLOP, the global and Local Penalty, a framework for capturing predictive heterogeneity and identifying predictive outliers. gLOP is based on penalized regression for multitask learning, which improves learning by leveraging training signal information from related tasks. We give two optimization algorithms for gLOP, one space-efficient, and another giving the full regularization path. We also characterize uniqueness in terms of the data and tuning parameters, and present empirical results on synthetic data and on two health research problems.} }
Endnote
%0 Conference Paper %T gLOP: the global and Local Penalty for Capturing Predictive Heterogeneity %A Rhiannon Rose %A Daniel Lizotte %B Proceedings of the 1st Machine Learning for Healthcare Conference %C Proceedings of Machine Learning Research %D 2016 %E Finale Doshi-Velez %E Jim Fackler %E David Kale %E Byron Wallace %E Jenna Wiens %F pmlr-v56-Rose16 %I PMLR %P 134--149 %U https://proceedings.mlr.press/v56/Rose16.html %V 56 %X When faced with a supervised learning problem, we hope to have rich enough data to build a model that predicts future instances well. However, in practice, problems can exhibit predictive heterogeneity: most instances might be relatively easy to predict, while others might be predictive outliers for which a model trained on the entire dataset does not perform well. Identifying these can help focus future data collection. We present gLOP, the global and Local Penalty, a framework for capturing predictive heterogeneity and identifying predictive outliers. gLOP is based on penalized regression for multitask learning, which improves learning by leveraging training signal information from related tasks. We give two optimization algorithms for gLOP, one space-efficient, and another giving the full regularization path. We also characterize uniqueness in terms of the data and tuning parameters, and present empirical results on synthetic data and on two health research problems.
RIS
TY - CPAPER TI - gLOP: the global and Local Penalty for Capturing Predictive Heterogeneity AU - Rhiannon Rose AU - Daniel Lizotte BT - Proceedings of the 1st Machine Learning for Healthcare Conference DA - 2016/12/10 ED - Finale Doshi-Velez ED - Jim Fackler ED - David Kale ED - Byron Wallace ED - Jenna Wiens ID - pmlr-v56-Rose16 PB - PMLR DP - Proceedings of Machine Learning Research VL - 56 SP - 134 EP - 149 L1 - http://proceedings.mlr.press/v56/Rose16.pdf UR - https://proceedings.mlr.press/v56/Rose16.html AB - When faced with a supervised learning problem, we hope to have rich enough data to build a model that predicts future instances well. However, in practice, problems can exhibit predictive heterogeneity: most instances might be relatively easy to predict, while others might be predictive outliers for which a model trained on the entire dataset does not perform well. Identifying these can help focus future data collection. We present gLOP, the global and Local Penalty, a framework for capturing predictive heterogeneity and identifying predictive outliers. gLOP is based on penalized regression for multitask learning, which improves learning by leveraging training signal information from related tasks. We give two optimization algorithms for gLOP, one space-efficient, and another giving the full regularization path. We also characterize uniqueness in terms of the data and tuning parameters, and present empirical results on synthetic data and on two health research problems. ER -
APA
Rose, R. & Lizotte, D.. (2016). gLOP: the global and Local Penalty for Capturing Predictive Heterogeneity. Proceedings of the 1st Machine Learning for Healthcare Conference, in Proceedings of Machine Learning Research 56:134-149 Available from https://proceedings.mlr.press/v56/Rose16.html.

Related Material