Abstract
This paper presents a rate-distortion theory for hierarchically networked data structures modelled as tree-indexed multitype process. Specifically, this paper gives a generalized asymptotic equipartition property (AEP) for hierarchical data structures. The proof of our main result uses large deviation principles for suitably defined empirical measures of multitype Galton-Watson trees.
Original language | English |
---|---|
Pages (from-to) | 1013-1024 |
Number of pages | 12 |
Journal | Far East Journal of Mathematical Sciences |
Volume | 101 |
Issue number | 5 |
DOIs | |
Publication status | Published - Mar 2017 |
Keywords
- Asymptotic equipartition property
- Empirical measure
- Perron-Frobenius eigenvalue
- Perron-Frobenius eigenvector
- Rate-distortion theory
- Relative entropy
- Weak irreducibility