A Comparative Analysis on the Performance of Upper Confidence Bound Algorithms in Adaptive Deep Neural Networks

Grigorios Papanikolaou, Ioannis Kontopoulos, Konstantinos Tserpes · 2026

Read A Comparative Analysis on the Performance of Upper Confidence Bound Algorithms in Adaptive Deep Neural Networks by Grigorios Papanikolaou, Ioannis Kontopoulos, Konstantinos Tserpes (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2604-24810.txt.

Description

Papanikolaou, Kontopoulos and Tserpes compare Upper Confidence Bound strategies for adaptive deep networks on edge devices. Adaptive networks use the bandit framework to pick thresholds for early exits without losing accuracy. Prior work used only the first Upper Confidence Bound strategy. The authors add four more, UCB-V, UCB-Tuned, UCB-Bayes and UCB-BwK, and test them on ResNet and MobileViT on CIFAR benchmarks. All achieve sub-linear regret, with UCB-Bayes fastest, while UCB-V and UCB-Tuned lead accuracy-latency and accuracy-energy trade-offs.

Catalog record

CREATORGrigorios Papanikolaou, Ioannis Kontopoulos, Konstantinos Tserpes
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE34130
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2604.24810

Browse the full library · About the Tower of Babel