Integrated Model for Informal Inference Based on Neural Networks

Authors:
Kyung-Joong Kim;Sung-Bae Cho
Affiliations:
Department of Computer Science, Yonsei University, Seoul, South Korea 120-749;Department of Computer Science, Yonsei University, Seoul, South Korea 120-749
Venue:
Neural Information Processing
Year:
2008

Citing 6
Cited 0

Probabilistic reasoning in intelligent systems: networks of plausible inference

Probabilistic reasoning in intelligent systems: networks of plausible inference
Self-organizing map with dynamical node splitting: application to handwritten digit recognition

Neural Computation
Soft combination of neural classifiers: a comparative study

Pattern Recognition Letters
Connectionist inference models

Neural Networks
Learning and inference in the brain

Neural Networks - Special issue: Neuroinformatics
A unified architecture for agent behaviors with selection of evolved neural network modules

Applied Intelligence

Quantified Score

Hi-index	0.00

Visualization

Abstract

Inference is one of human's high-level functionalities and it is not easy to implement in machine. It is believed that inference is not results of single neuron's activity. Instead, it is a complex activity generated by multiple neural networks. Unlike computer, it is more flexible and concludes differently even for the similar situations in case of human. In this paper, these characteristics are defined as "informality." Informality in inference can be implemented using the interaction of multiple neural networks with the inclusion of internal or subjective properties. Simple inference tasks such as pattern recognition and robot control are solved based on the informal inference ideas. Especially, fuzzy integral and behavior network methods are adopted to realize that. Experimental results show that the informal inference can perform better with more flexibility compared to the previous static approaches.