Abstract
The thesis studies three important applications of random matrices to information processing. Our main contribution is that we consider probabilistic
systems involving more general random matrix ensembles than the classical ensembles with iid entries, i.e. models that account for statistical dependence
between the entries. Specifically, the involved matrices are invariant or fulfill a certain asymptotic freeness condition as their dimensions grow to infinity.
Informally speaking, all latent variables contribute to the system model in a democratic fashion – there are no preferred latent variables in the system.
systems involving more general random matrix ensembles than the classical ensembles with iid entries, i.e. models that account for statistical dependence
between the entries. Specifically, the involved matrices are invariant or fulfill a certain asymptotic freeness condition as their dimensions grow to infinity.
Informally speaking, all latent variables contribute to the system model in a democratic fashion – there are no preferred latent variables in the system.
| Originalsprog | Engelsk |
|---|---|
| Vejledere |
|
| Udgiver | |
| ISBN'er, elektronisk | 978-87-7112-828-4 |
| DOI | |
| Status | Udgivet - 2016 |
Bibliografisk note
PhD supervisor:Professor Bernard Henri Fleury, Aalborg University, Demmark