An Information Bottleneck Problem with Rényi's Entropy
arXiv:2101.12564
Abstract
This paper considers an information bottleneck problem with the objective of obtaining a most informative representation of a hidden feature subject to a Rényi entropy complexity constraint. The optimal bottleneck trade-off between relevance (measured via Shannon's mutual information) and Rényi entropy cost is defined and an iterative algorithm for finding approximate solutions is provided. We also derive an operational characterization for the optimal trade-off by demonstrating that the optimal Rényi entropy-relevance trade-off is achievable by a simple time-sharing scalar coding scheme and that no coding scheme can provide better performance. Two examples where the optimal Shannon entropy-relevance trade-off can be exactly determined are further given.
arXiv comment5 pages, 4 figures, submitted to ISIT 2021
References in corpus (7)
- The information bottleneck method
- Rényi Divergence and Kullback-Leibler Divergence
- Opening the Black Box of Deep Neural Networks via Information
- Deep Variational Information Bottleneck
- Learning Representations for Neural Network-Based Classification Using the Information Bottleneck Principle
- Rényi Bounds on Information Combining
- Hard Clusters Maximize Mutual Information