Curiosity-Driven Exploration with Information Bottleneck Representations and Matrix-Based Mutual Information
Entropy · 02.06.2026 · 28 (6) · 625
DOI: 10.3390/e28060625 ↗Kurz gefasst
Fachartikel in der Zeitschrift Entropy, online erschienen am 02.06.2026. Verfasst von Zhaoxu Meng und 1 weiteren Autor:innen. Beteiligte Einrichtung: Department of Computer Science, The University of Hong Kong, Pokfulam, Hong Kong, China. Die Arbeit nennt 49 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Key Science and Technology Project of the Ministry of Emergency Management of the People's Republic of China, Key Science and Technology Project of the Ministry of Emergency Management of the People’s Republic of China.
Abstract (Originalsprache)
Curiosity empowers humans to ask questions about the world and explore it without relying on extrinsic, encouraging rewards such as money. To investigate how this mechanism drives exploration, we implement a curiosity-based approach and test it in a reinforcement learning environment. We define curiosity using a hybrid intrinsic signal based on prediction error and the rarity of state–action pairs. To address the curse of dimensionality in raw pixel inputs, we adopt the Information Bottleneck (IB) principle to learn low-dimensional representations that are both compact and predictive. We introduce two formulations for computing mutual information—one based on entropy decomposition and the other on matrix-based Rényi entropy—and compare their effectiveness. Experiments on Acrobot show substantially improved exploration efficiency over Intrinsic Curiosity Module (ICM), Random Network Distillation (RND), and a k-NN novelty baseline, while results on MountainCar indicate that the proposed method is not uniformly superior in low-dimensional environments. These findings suggest that IB-shaped representations and matrix-based information objectives are most beneficial when observations are high-dimensional or dynamics are structurally complex.
Schlagwörter
Stichwörter der Autor:innen
Autor:innen und Institutionen
- Department of Computer Science, The University of Hong Kong, Pokfulam, Hong Kong, China.
- School of Automation Science and Electrical Engineering, Beihang University, No. 37 Xueyuan Road, Haidian District, Beijing 100191, China.
ORCID-Profile: Zhaoxu Meng, Yong Cui
Förderung
- Key Science and Technology Project of the Ministry of Emergency Management of the People's Republic of China
- Key Science and Technology Project of the Ministry of Emergency Management of the People’s Republic of China
Zitieren
APA
Meng, Z., & Cui, Y. (2026). Curiosity-Driven Exploration with Information Bottleneck Representations and Matrix-Based Mutual Information. Entropy, 28(6), 625. https://doi.org/10.3390/e28060625
Vancouver
Meng Z, Cui Y. Curiosity-Driven Exploration with Information Bottleneck Representations and Matrix-Based Mutual Information. Entropy. 2026;28(6):625. doi:10.3390/e28060625
BibTeX
@article{Meng2026,
author = {Meng, Zhaoxu and Cui, Yong},
title = {Curiosity-Driven Exploration with Information Bottleneck Representations and Matrix-Based Mutual Information},
journal = {Entropy},
year = {2026},
volume = {28},
number = {6},
pages = {625},
doi = {10.3390/e28060625}
}
EUVIMED