Generating Functionals for Computational Intelligence: The Fisher Information as an Objective Function for Self-Limiting Hebbian Learning Rules.
Rodrigo EchevesteClaudius GrosPublished in: Frontiers Robotics AI (2014)
Keyphrases
- learning rules
- fisher information
- objective function
- artificial neural networks
- biologically inspired
- neural network
- information geometry
- learning algorithm
- probability density function
- optimal solution
- euclidean space
- back propagation
- positive definite
- noise level
- lower bound
- estimation error
- learning scheme
- geodesic distance
- fuzzy logic
- multidimensional scaling
- exponential family
- special case
- covariance matrix
- shape analysis
- feature space
- kl divergence
- image processing