An Optimal Policy for Target Localization with Application to Electron Microscopy

Raphael Sznitman, Aurelien Lucchi, Peter Frazier, Bruno Jedynak, Pascal Fua
Proceedings of the 30th International Conference on Machine Learning, PMLR 28(1):1-9, 2013.

Abstract

This paper considers the task of finding a target location by making a limited number of sequential observations. Each observation results from evaluating an imperfect classifier of a chosen cost and accuracy on an interval of chosen length and position. Within a Bayesian framework, we study the problem of minimizing an objective that combines the entropy of the posterior distribution with the cost of the questions asked. In this problem, we show that the one-step lookahead policy is Bayes-optimal for any arbitrary time horizon. Moreover, this one-step lookahead policy is easy to compute and implement. We then use this policy in the context of localizing mitochondria in electron microscope images, and experimentally show that significant speed ups in acquisition can be gained, while maintaining near equal image quality at target locations, when compared to current policies.

Cite this Paper


BibTeX
@InProceedings{pmlr-v28-sznitman13, title = {An Optimal Policy for Target Localization with Application to Electron Microscopy}, author = {Sznitman, Raphael and Lucchi, Aurelien and Frazier, Peter and Jedynak, Bruno and Fua, Pascal}, booktitle = {Proceedings of the 30th International Conference on Machine Learning}, pages = {1--9}, year = {2013}, editor = {Dasgupta, Sanjoy and McAllester, David}, volume = {28}, number = {1}, series = {Proceedings of Machine Learning Research}, address = {Atlanta, Georgia, USA}, month = {17--19 Jun}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v28/sznitman13.pdf}, url = {https://proceedings.mlr.press/v28/sznitman13.html}, abstract = {This paper considers the task of finding a target location by making a limited number of sequential observations. Each observation results from evaluating an imperfect classifier of a chosen cost and accuracy on an interval of chosen length and position. Within a Bayesian framework, we study the problem of minimizing an objective that combines the entropy of the posterior distribution with the cost of the questions asked. In this problem, we show that the one-step lookahead policy is Bayes-optimal for any arbitrary time horizon. Moreover, this one-step lookahead policy is easy to compute and implement. We then use this policy in the context of localizing mitochondria in electron microscope images, and experimentally show that significant speed ups in acquisition can be gained, while maintaining near equal image quality at target locations, when compared to current policies.} }
Endnote
%0 Conference Paper %T An Optimal Policy for Target Localization with Application to Electron Microscopy %A Raphael Sznitman %A Aurelien Lucchi %A Peter Frazier %A Bruno Jedynak %A Pascal Fua %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-sznitman13 %I PMLR %P 1--9 %U https://proceedings.mlr.press/v28/sznitman13.html %V 28 %N 1 %X This paper considers the task of finding a target location by making a limited number of sequential observations. Each observation results from evaluating an imperfect classifier of a chosen cost and accuracy on an interval of chosen length and position. Within a Bayesian framework, we study the problem of minimizing an objective that combines the entropy of the posterior distribution with the cost of the questions asked. In this problem, we show that the one-step lookahead policy is Bayes-optimal for any arbitrary time horizon. Moreover, this one-step lookahead policy is easy to compute and implement. We then use this policy in the context of localizing mitochondria in electron microscope images, and experimentally show that significant speed ups in acquisition can be gained, while maintaining near equal image quality at target locations, when compared to current policies.
RIS
TY - CPAPER TI - An Optimal Policy for Target Localization with Application to Electron Microscopy AU - Raphael Sznitman AU - Aurelien Lucchi AU - Peter Frazier AU - Bruno Jedynak AU - Pascal Fua BT - Proceedings of the 30th International Conference on Machine Learning DA - 2013/02/13 ED - Sanjoy Dasgupta ED - David McAllester ID - pmlr-v28-sznitman13 PB - PMLR DP - Proceedings of Machine Learning Research VL - 28 IS - 1 SP - 1 EP - 9 L1 - http://proceedings.mlr.press/v28/sznitman13.pdf UR - https://proceedings.mlr.press/v28/sznitman13.html AB - This paper considers the task of finding a target location by making a limited number of sequential observations. Each observation results from evaluating an imperfect classifier of a chosen cost and accuracy on an interval of chosen length and position. Within a Bayesian framework, we study the problem of minimizing an objective that combines the entropy of the posterior distribution with the cost of the questions asked. In this problem, we show that the one-step lookahead policy is Bayes-optimal for any arbitrary time horizon. Moreover, this one-step lookahead policy is easy to compute and implement. We then use this policy in the context of localizing mitochondria in electron microscope images, and experimentally show that significant speed ups in acquisition can be gained, while maintaining near equal image quality at target locations, when compared to current policies. ER -
APA
Sznitman, R., Lucchi, A., Frazier, P., Jedynak, B. & Fua, P.. (2013). An Optimal Policy for Target Localization with Application to Electron Microscopy. Proceedings of the 30th International Conference on Machine Learning, in Proceedings of Machine Learning Research 28(1):1-9 Available from https://proceedings.mlr.press/v28/sznitman13.html.

Related Material