Additive hybrid networks for fuzzy logic

James M. Keller, Yoichi Hayashi, Zhihong Chen

Research output: Contribution to journalArticle

3 Citations (Scopus)


Evidence aggregation networks based on multiplicative fuzzy hybrid operators were introduced by Krishnapuram and Lee. They have been used for image segmentation, pattern recognition, and general multicriteria decision-making. One of the drawbacks of these networks is that the training is complex and quite time consuming. We have recently modified these aggregation networks to implement additive fuzzy hybrid connectives. These new networks have similar excellent properties for decision-making under uncertainty as do their multiplicative precursors, and have the advantage that training is easier since the additive operators are not as complex in form. In earlier work, trainable neural networks have been used to perform inference in a fuzzy logic system. Fixed-architecture networks of fuzzy operators and standard neural networks have been utilized. The additive hybrid operators are extremely flexible and so, are excellent candidates for nodes in a network structure for fuzzy logic inference. The purpose of this paper is to demonstrate the capability of networks of additive hybrid nodes to learn appropriate functional relationships for fuzzy logic. An added advantage of these networks is that they are transparent, i.e., after training, the individual nodes can be analyzed as a collection of "mini-rules".

Original languageEnglish
Pages (from-to)307-313
Number of pages7
JournalFuzzy Sets and Systems
Issue number3
Publication statusPublished - 26 Sep 1994


  • Additive hybrid operators
  • Approximate reasoning
  • Fuzzy logic inference
  • Fuzzy neurons
  • Neural networks

Fingerprint Dive into the research topics of 'Additive hybrid networks for fuzzy logic'. Together they form a unique fingerprint.

  • Cite this