Dynamics Aware Past-Future Information Bottleneck
Abstract
The simplification of complex dynamical systems is essential for efficient estimation and control. A principled approach to this task is through the lens of information-theoretic compression. The Past-Future Information Bottleneck (PF-IB) provides a theoretical framework for compressing linear dynamical systems. However, the standard formulation often lacks structural consistency, yielding representations that are not realizable as dynamical systems. By constraining the objective to a specific manifold of block Hankel matrices, which we prove is a Lie group, we bridge the gap between PF-IB and realization theory. Our approach ensures that the resulting compressed dynamics are realizable, thereby enabling its direct application to estimation and control. We demonstrate our approach in numerical simulations.
Related
- Matrix Triangle Inequalities: Past, Present, and Future
- On the Information Bottleneck for Stable Variables
- Optimal Multitask Representations Under a Rate Constraint: An Information Bottleneck Analysis
- On the Cardinality of Optimal Representations in the Binary-Source Information Bottleneck
- Training-Adaptive Convolutional Sparse Coding via Information Bottleneck for Robust Visual Representation
BibTeX
@misc{chung2026dynamics,
title = {{Dynamics Aware Past-Future Information Bottleneck}},
author = {Wooyoung Chung and Stas Tiomkin},
howpublished = {arXiv},
year = {2026},
month = nov,
doi = {10.48550/arxiv.2611.08426},
eprint = {2611.08426},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2611.08426},
}
RIS
TY - UNPB
TI - Dynamics Aware Past-Future Information Bottleneck
AU - Wooyoung Chung
AU - Stas Tiomkin
PY - 2026
DA - 2026/11/05
PB - arXiv
DO - 10.48550/arxiv.2611.08426
UR - https://arxiv.org/abs/2611.08426
AB - The simplification of complex dynamical systems is essential for efficient estimation and control. A principled approach to this task is through the lens of information-theoretic compression. The Past-Future Information Bottleneck (PF-IB) provides a theoretical framework for compressing linear dynamical systems. However, the standard formulation often lacks structural consistency, yielding representations that are not realizable as dynamical systems. By constraining the objective to a specific manifold of block Hankel matrices, which we prove is a Lie group, we bridge the gap between PF-IB and realization theory. Our approach ensures that the resulting compressed dynamics are realizable, thereby enabling its direct application to estimation and control. We demonstrate our approach in numerical simulations.
ER -