A Neural Network Diagnosis Model without Disorder Independence Assumption

Yong Xu, Zhang, C

Abstract

. Generally, the disorders in a neural network diagnosis model are assumed independent each other. In this paper, we propose a neural network model for diagnostic problem solving where the disorder independence assumption is no longer necessary. Firstly, we characterize the diagnostic tasks and the causal network which is used to represent the diagnostic problem, then we describe the neural network diagnosis model, finally, some experiment results will be given. 1 Introduction Finding explanations for a given set of events is an important aspect of general intelligent behaviour. The process of finding the best explanation was defined as Abduction by the philosopher C. S. Peirce [8]. Diagnosis is a typical abductive problem. For a set of manifestations(observations), the diagnostic inference is to find the most plausible faults or disorders which can explain the manifestations observed. In general, an individual fault or disorder can explain only a portion of the manifestations. Ther...
BibTeX
@inproceedings{xu1998neural,
  author = {Yong Xu and Zhang, C},
  title = {{A Neural Network Diagnosis Model without Disorder Independence Assumption}},
  booktitle = {PRICAI},
  year = {1998},
  doi = {10.1007/bfb0095282},
}

Loading related content...

Figure