Optimal Allocation of Information Granularity to the Inputs of Granular Neural Networks

• Author / Creator

  Akhoundi, Elaheh

• In this thesis, we propose a design process to construct granular neural networks with granular inputs and numeric network parameters. The proposed granular network is formed on the basis of a numeric neural network whose inputs are augmented using probabilistic information granules. The design problem is formulated as an optimization problem which aims to allocate a given level of information granularity to the inputs of the network such that the specificity of the network outputs gets maximized. The resulting optimization problem is solved analytically and the derived solution determines the optimal granularity levels corresponding to the input features of the granular neural network. The proposed design process is then used to construct granular neural networks for several synthetic and real data sets.

• Subjects / Keywords

  • Sensitivity Analysis
  • Granular Computing
  • Optimal Granularity Allocation
  • Granular Neural Networks

• Graduation date

  Fall 2014

• Type of Item

  Thesis

• Degree

  Master of Science

• DOI

  https://doi.org/10.7939/R3J960J49

• License

  This thesis is made available by the University of Alberta Libraries with permission of the copyright owner solely for non-commercial purposes. This thesis, or any portion thereof, may not otherwise be copied or reproduced without the written consent of the copyright owner, except to the extent permitted by Canadian copyright law.