Inhalt

•  
   Contents
•  
   Introduction
•  
  Problem Statement
•  
   Robust Processing in Human-Computer Interaction
•  
   Basic Principles of Computer Vision
•  
   Basic Principles of Automatic Speech Understanding
•  
  Integration of Speech and Image Processing -- An Overview
•  
   Psychological experiments and the level of information processing
•  
   Linguistics and the symbol grounding problem
•  
   Spatial cognition
•  
   A categorization of computational systems
•  
   The Correspondence Problem
•  
   Knowledge representation and control structures
•  
   Spatial models
•  
   Learning
•  
   Other Related Work
•  
  Contributions
•  
   A probabilistic translation scheme
•  
   A separate integration and interaction component for speech understanding and vision base-line systems
•  
   The choice of the application area
•  
   Inference and learning
•  
   A Model for Uncertainty
•  
   Intensional and extensional models
•  
   Bayesian Networks
•  
   Definition of Bayesian networks
•  
   Modeling in Bayesian networks
•  
   How to get those numbers? Some simplification
•  
   Modeling corresponding variables
•  
   Inference in Bayesian Networks
•  
   I-maps, moral graphs, and d-separation
•  
   Singly connected networks
•  
   Coping with loops
•  
   A conditional bucket elimination scheme
•  
   Relation to Graph Matching
•  
   Applications of Bayesian Networks
•  
  Bayesian networks for integration of speech and images
•  
   Modeling principles
•  
   Inference methods
•  
   An application to human-computer interaction
•  
  Modeling
•  
   Scenario and Domain Description
•  
   Experimental Data
•  
   The General System Architecture
•  
   The speech understanding and dialog components
•  
   The object recognition component
•  
   Speech understanding and vision results
•  
  Spatial Modeling
•  
   A model for 3-d projective relations
•  
   The spatial model in two dimensions
•  
   The neighborhood graph
•  
   Localization attributes
•  
   Summary
•  
  Object Identification using Bayesian Networks
•  
   Previous work
•  
   Starting points for improvements
•  
   An extended Bayesian model for object classes
•  
   A Bayesian model for spatial relations
•  
   Modeling structural relationships
•  
   Integrating the what and where
•  
   Summary
•  
  Inference and Learning
•  
   Establishing referential links
•  
  Interaction of speech and image understanding
•  
   The most probable class of the intended object
•  
   Interpretation of structural descriptions
•  
   Unknown object names
•  
   Disambiguating alternative interpretations of an utterance
•  
   Disambiguating the selected reference frame
•  
   Detection of neighborhood relations
•  
   Further Learning Capabilities
•  
  Results
•  
   Test Sets
•  
   Classification of System Answers
•  
   Results on the Select-Obj test set
•  
  Results on the Select-Rel test set
•  
   Verification of the neighborhood assumption
•  
   Identification results
•  
   Qualitative results
•  
   Object Classification using Speech and Image Features
•  
   Summary
•  
  Summary and Conclusion
•  
   The Integration of Speech and Images as a Probabilistic Decoding Process
•  
   Contributions
•  
   Future Work
•  
   Final Remarks
•  
   The elementary objects of baufix210
•  
   Bibliography
•  
   Index