Contents

• Abstract
• Preface
• Contents
• List of abbreviations and glossary of terms
• Introduction
  • Contributions and structure of the thesis
• Publications of the thesis
  • Original papers
  • Contents of the publications and author's contributions
• Bayesian probability theory
  • Propositions
  • Elementary rules of Bayesian probability theory
    • Bayes' rule
    • Marginalisation principle
  • Decision theory
  • Summary: learning, reasoning and action
• Bayesian learning in practice
  • Probability density for real valued variables
  • Methods for approximating the posterior probability
    • Point estimates
      • EM algorithm.
    • Stochastic sampling
    • Parametric approximations
  • Information-theoretic approaches to learning
    • Coding and complexity
    • Minimum message length inference
  • Ensemble learning
    • Cost function
    • Bits-back argument
  • Specification of the model and priors
    • Noise models
    • Causal structure of the model
    • Supervised vs. unsupervised learning
    • Priors
      • Hierarchical model instead of prior.
      • Uninformative priors.
• Linear factor analysis and its extensions
  • Linear Gaussian factor analysis model
    • Neural network interpretation of the model
    • Algorithms
  • Non-Gaussian factors
    • Algorithms
  • Nonlinear mapping
    • Multi-layer perceptron networks
    • Algorithms
      • Self-organising map
      • Auto-associative models.
  • Dynamic factors
    • Algorithms
• Bayesian nonlinear factor analysis
  • Model
  • Why simple methods fail
  • Approximation of the posterior
  • Automatic pruning
  • Results
• Discussion
  • Biological relevance
    • Cerebral cortex and generative models
    • Cortex and nonlinear factor analysis
    • Structural development
  • Future trends
• References
• Publication I
• Publication II
• Publication III
• Publication IV
• Publication V
• Publication VI
• Publication VII
• Publication VIII