A proof of concept for enhancing branchynet with trainable reference-vector-based early exits
Discover Artificial Intelligence · 17.11.2026 · 6 (1)
DOI: 10.1007/s44163-026-02676-0 ↗Kurz gefasst
Fachartikel in der Zeitschrift Discover Artificial Intelligence, online erschienen am 17.11.2026. Verfasst von Thomas Prantl und 5 weiteren Autor:innen. Die Arbeit nennt 25 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Julius-Maximilians-Universität Würzburg.
Abstract (Originalsprache)
As machine learning research advances, models continue to grow in both quality and scale. This trend increases computational costs and complicates large-scale deployment. Although various techniques reduce model size or accelerate inference with minimal accuracy loss, they are often complex to use and require substantial structural modifications. We present RefNet, a lightweight early exit framework that replaces BranchyNet’s computationally heavier auxiliary classifier heads with learnable, class-wise reference vectors. At each exit, RefNet computes the Euclidean distance between an intermediate feature representation and a small set of trainable per-class reference vectors and uses the resulting similarity scores both for classification and for the early-exit decision. In a proof-of-concept study using a ResNet-50 backbone on CIFAR-10, CIFAR-100, SVHN, and STL-10, with results averaged over five independent training and evaluation runs for each configuration, RefNet achieves inference speedups comparable to or greater than those of BranchyNet, while incurring lower training and parameter overhead: on average, it increases the parameter count by only 0.2% (vs. 1.7% for BranchyNet) and the training time by 1.1% (vs. 2.3% for BranchyNet). At its default operating point, using two early exits with confidence thresholds of 70% and 80%, respectively, RefNet achieves an inference speedup of up to 1.67 $$\times $$ on SVHN, with a mean accuracy reduction of only 0.14 percentage points relative to the vanilla ResNet-50. More aggressive thresholds increase the speedup on SVHN to 1.70 $$\times $$ at a correspondingly larger but still moderate accuracy cost. Welch’s t -tests over the five runs show that RefNet’s reduction in inference time relative to BranchyNet is statistically significant on all four datasets ( $$p\le 0.001$$ ), whereas the SVHN accuracy difference to the vanilla ResNet-50 at the default operating point is not ( $$p=0.25$$ ). These results demonstrate the feasibility of similarity-based early exit decision making as a lightweight alternative to learned classifier heads. RefNet does not cache or reuse computation across inference requests.
Förderung
- Julius-Maximilians-Universität Würzburg
Zitieren
APA
Prantl, T., Ginzel, P., Borst, V., Horn, L., Engel, S., & Kounev, S. (2026). A proof of concept for enhancing branchynet with trainable reference-vector-based early exits. Discover Artificial Intelligence, 6(1). https://doi.org/10.1007/s44163-026-02676-0
Vancouver
Prantl T, Ginzel P, Borst V, Horn L, Engel S, Kounev S. A proof of concept for enhancing branchynet with trainable reference-vector-based early exits. Discover Artificial Intelligence. 2026;6(1). doi:10.1007/s44163-026-02676-0
BibTeX
@article{Prantl2026,
author = {Prantl, Thomas and Ginzel, Phillip and Borst, Vanessa and Horn, Lukas and Engel, Simon and Kounev, Samuel},
title = {A proof of concept for enhancing branchynet with trainable reference-vector-based early exits},
journal = {Discover Artificial Intelligence},
year = {2026},
volume = {6},
number = {1},
doi = {10.1007/s44163-026-02676-0}
}
EUVIMED