Theoretical studies and practical applications of recurrent neural networks based on architectural biasSupervisor: doc. Ing. Michal Čerňanský, PhD.
This page shows details on the project. The primary projects are displayed together with a list of sub-projects.
|Project description:||Markovian architectural bias is a property realeated with a fixed point attractor state space dynamics of architectures with recurrent connections. Models such as neural prediction machine and fractal prediction machine successfully use this type of dynamics, as well as the novel Echo State Networks with huge randomly interconnected hidden layer. We study the network´s state space dynamics and its possible adjustment so some specific sequences can be successfully processed by proposed methods. We will specify practical problem domains (for example image or sound processing), where proposed approaches can be used.|
|Kind of project:||APVV ()|
|Department:||Institute of Applied Informatics (FIIT)|
|Project status:||Successfully completed|
|Project start date :||01. 01. 2005|
|Project close date:||31. 12. 2007|
|Number of workers in the project:||1|
|Number of official workers in the project:||0|