On the estimation of differential entropy for Gaussian mixtures
Summary
No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.
Summarise with VIRAT Free to try with your Science Catalogs account.Abstract
We develop a novel method for approximating the differential entropy of Gaussian mixtures that remains accurate in high-dimensional settings. The proposed approach overcomes the numerical and analytical challenges that typically arise as dimensionality increases, making it suitable for modern high-dimensional data applications. In addition, we address the fingerprinting (collusion-resistant watermarking) problem in the context of multiple TV channels. We formulate and solve the associated information-theoretic model, providing asymptotic optimal strategies under realistic attack scenarios.
Fields of Study & Subjects
Bayesian Methods and Mixture Models
Advanced Steganography and Watermarking Techniques
Probability and Statistical Research
Publication details
- Journal / source
- TU/e Research Portal
- Publication date
- 12 March 2026
- Language
- EN