Fisher information and α-connections for a class of transformational models
✍ Scribed by Marco A.N. Fernandes; Luiz A.B. San Martin
- Book ID
- 104358075
- Publisher
- Elsevier Science
- Year
- 2000
- Tongue
- English
- Weight
- 132 KB
- Volume
- 12
- Category
- Article
- ISSN
- 0926-2245
No coin nor oath required. For personal study only.
✦ Synopsis
In this paper we use invariant theory for representations of groups in order to get an indirect method of computing the Fisher metric and the α-connections of transformational models parameterized by symmetric spaces. Among others these models include the Von Mises-Fisher model, the Hyperboloid model, the multivariable zero mean normal model with determinant one covariant matrix and the Wishard model.
📜 SIMILAR VOLUMES
In this paper we deduce a class of canonical models for weakly connected neural networks depending on first and second order adaptation conditions. An adaptation condition is a relation involving internal and external network parameters that translates the network's adjustment to environmental stimu