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

Florian Valade, Mohamed Hebiri, Paul Gay

Keywords: Early Exit, Reject Option, Classification

TL;DR: We propose a method to improve early exit networks by taking into account Risk of each early exit and respecting a specified budget.

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.

Code Link: https://github.com/FlorianVal/Early-Exit-With-Reject-Option

Readers: auai.org/UAI/2025/Conference, auai.org/UAI/2025/Conference/Area_Chairs, auai.org/UAI/2025/Conference/Reviewers, auai.org/UAI/2025/Conference/Submission390/Authors, auai.org/UAI/2025/Conference/Submission390/Reproducibility_Reviewers

Submission Number: 390