Dynamic Sensing: Better Classification under Acquisition Constraints

Oran Richman, Shie Mannor
Proceedings of the 32nd International Conference on Machine Learning, PMLR 37:267-275, 2015.

Abstract

In many machine learning applications the quality of the data is limited by resource constraints (may it be power, bandwidth, memory, ...). In such cases, the constraints are on the average resources allocated, therefore there is some control over each sample’s quality. In most cases this option remains unused and the data’s quality is uniform over the samples. In this paper we propose to actively allocate resources to each sample such that resources are used optimally overall. We propose a method to compute the optimal resource allocation. We further derive generalization bounds for the case where the problem’s model is unknown. We demonstrate the potential benefit of this approach on both simulated and real-life problems.

Cite this Paper

BibTeX
@InProceedings{pmlr-v37-richman15, title = {Dynamic Sensing: Better Classification under Acquisition Constraints}, author = {Richman, Oran and Mannor, Shie}, booktitle = {Proceedings of the 32nd International Conference on Machine Learning}, pages = {267--275}, 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/richman15.pdf}, url = {https://proceedings.mlr.press/v37/richman15.html}, abstract = {In many machine learning applications the quality of the data is limited by resource constraints (may it be power, bandwidth, memory, ...). In such cases, the constraints are on the average resources allocated, therefore there is some control over each sample’s quality. In most cases this option remains unused and the data’s quality is uniform over the samples. In this paper we propose to actively allocate resources to each sample such that resources are used optimally overall. We propose a method to compute the optimal resource allocation. We further derive generalization bounds for the case where the problem’s model is unknown. We demonstrate the potential benefit of this approach on both simulated and real-life problems.} }
Endnote
%0 Conference Paper %T Dynamic Sensing: Better Classification under Acquisition Constraints %A Oran Richman %A Shie Mannor %B Proceedings of the 32nd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2015 %E Francis Bach %E David Blei %F pmlr-v37-richman15 %I PMLR %P 267--275 %U https://proceedings.mlr.press/v37/richman15.html %V 37 %X In many machine learning applications the quality of the data is limited by resource constraints (may it be power, bandwidth, memory, ...). In such cases, the constraints are on the average resources allocated, therefore there is some control over each sample’s quality. In most cases this option remains unused and the data’s quality is uniform over the samples. In this paper we propose to actively allocate resources to each sample such that resources are used optimally overall. We propose a method to compute the optimal resource allocation. We further derive generalization bounds for the case where the problem’s model is unknown. We demonstrate the potential benefit of this approach on both simulated and real-life problems.
RIS
TY - CPAPER TI - Dynamic Sensing: Better Classification under Acquisition Constraints AU - Oran Richman AU - Shie Mannor BT - Proceedings of the 32nd International Conference on Machine Learning DA - 2015/06/01 ED - Francis Bach ED - David Blei ID - pmlr-v37-richman15 PB - PMLR DP - Proceedings of Machine Learning Research VL - 37 SP - 267 EP - 275 L1 - http://proceedings.mlr.press/v37/richman15.pdf UR - https://proceedings.mlr.press/v37/richman15.html AB - In many machine learning applications the quality of the data is limited by resource constraints (may it be power, bandwidth, memory, ...). In such cases, the constraints are on the average resources allocated, therefore there is some control over each sample’s quality. In most cases this option remains unused and the data’s quality is uniform over the samples. In this paper we propose to actively allocate resources to each sample such that resources are used optimally overall. We propose a method to compute the optimal resource allocation. We further derive generalization bounds for the case where the problem’s model is unknown. We demonstrate the potential benefit of this approach on both simulated and real-life problems. ER -
APA
Richman, O. & Mannor, S.. (2015). Dynamic Sensing: Better Classification under Acquisition Constraints. Proceedings of the 32nd International Conference on Machine Learning, in Proceedings of Machine Learning Research 37:267-275 Available from https://proceedings.mlr.press/v37/richman15.html.

Related Material