EERO: Early Exit with Reject Option for Efficient Classification with limited budget

Florian Valade, Mohamed Hebiri, Paul Gay
Proceedings of the Forty-first Conference on Uncertainty in Artificial Intelligence, PMLR 286:4290-4308, 2025.

Abstract

The increasing complexity of advanced machine learning models requires innovative approaches to manage computational resources effectively. One such method is the Early Exit strategy, which allows for adaptive computation by providing a mechanism to shorten the processing path for simpler data instances. In this paper, we propose EERO, a new methodology to translate the problem of early exiting to a problem of using multiple classifiers with reject option in order to better select the exiting head for each instance. We calibrate the probabilities of exiting at the different heads using aggregation with exponential weights to guarantee a fixed budget. We consider factors such as Bayesian risk, budget constraints, and head-specific budget consumption. Experimental results demonstrate that our method achieves competitive compromise between budget allocation and accuracy.

Cite this Paper

BibTeX
@InProceedings{pmlr-v286-valade25a, title = {EERO: Early Exit with Reject Option for Efficient Classification with limited budget}, author = {Valade, Florian and Hebiri, Mohamed and Gay, Paul}, booktitle = {Proceedings of the Forty-first Conference on Uncertainty in Artificial Intelligence}, pages = {4290--4308}, year = {2025}, editor = {Chiappa, Silvia and Magliacane, Sara}, volume = {286}, series = {Proceedings of Machine Learning Research}, month = {21--25 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v286/main/assets/valade25a/valade25a.pdf}, url = {https://proceedings.mlr.press/v286/valade25a.html}, abstract = {The increasing complexity of advanced machine learning models requires innovative approaches to manage computational resources effectively. One such method is the Early Exit strategy, which allows for adaptive computation by providing a mechanism to shorten the processing path for simpler data instances. In this paper, we propose EERO, a new methodology to translate the problem of early exiting to a problem of using multiple classifiers with reject option in order to better select the exiting head for each instance. We calibrate the probabilities of exiting at the different heads using aggregation with exponential weights to guarantee a fixed budget. We consider factors such as Bayesian risk, budget constraints, and head-specific budget consumption. Experimental results demonstrate that our method achieves competitive compromise between budget allocation and accuracy.} }
Endnote
%0 Conference Paper %T EERO: Early Exit with Reject Option for Efficient Classification with limited budget %A Florian Valade %A Mohamed Hebiri %A Paul Gay %B Proceedings of the Forty-first Conference on Uncertainty in Artificial Intelligence %C Proceedings of Machine Learning Research %D 2025 %E Silvia Chiappa %E Sara Magliacane %F pmlr-v286-valade25a %I PMLR %P 4290--4308 %U https://proceedings.mlr.press/v286/valade25a.html %V 286 %X The increasing complexity of advanced machine learning models requires innovative approaches to manage computational resources effectively. One such method is the Early Exit strategy, which allows for adaptive computation by providing a mechanism to shorten the processing path for simpler data instances. In this paper, we propose EERO, a new methodology to translate the problem of early exiting to a problem of using multiple classifiers with reject option in order to better select the exiting head for each instance. We calibrate the probabilities of exiting at the different heads using aggregation with exponential weights to guarantee a fixed budget. We consider factors such as Bayesian risk, budget constraints, and head-specific budget consumption. Experimental results demonstrate that our method achieves competitive compromise between budget allocation and accuracy.
APA
Valade, F., Hebiri, M. & Gay, P.. (2025). EERO: Early Exit with Reject Option for Efficient Classification with limited budget. Proceedings of the Forty-first Conference on Uncertainty in Artificial Intelligence, in Proceedings of Machine Learning Research 286:4290-4308 Available from https://proceedings.mlr.press/v286/valade25a.html.

Related Material