Rights statement: The final publication is available at Springer via http://dx.doi.org/10.1007/s12530-020-09336-3
Accepted author manuscript, 416 KB, PDF document
Available under license: CC BY-NC: Creative Commons Attribution-NonCommercial 4.0 International License
Final published version
Licence: None
Research output: Contribution to Journal/Magazine › Journal article › peer-review
Research output: Contribution to Journal/Magazine › Journal article › peer-review
}
TY - JOUR
T1 - Autonomous Data Density pruning fuzzy neural network for Optical Interconnection Network
AU - de Campos Souza, Paulo Vitor
AU - Soares, Eduardo A.
AU - Guimarães, Augusto Junio
AU - Araujo, Vanessa Souza
AU - Araujo, Vinicius Jonathan S.
AU - Rezende, Thiago Silva
N1 - The final publication is available at Springer via http://dx.doi.org/10.1007/s12530-020-09336-3
PY - 2021/12/31
Y1 - 2021/12/31
N2 - Traditionally, fuzzy neural networks have parametric clustering methods based on equally spaced membership functions to fuzzify inputs of the model. In this sense, it produces an excessive number calculations for the parameters’ definition of the network architecture, which may be a problem especially for real-time large-scale tasks. Therefore, this paper proposes a new model that uses a non-parametric technique for the fuzzification process. The proposed model uses an autonomous data density approach in a pruned fuzzy neural network, wich favours the compactness of the model. The performance of the proposed approach is evaluated through the usage of databases related to the Optical Interconnection Network. Finally, binary patterns classification tests for the identification of temporal distribution (asynchronous or client–server) were performed and compared with state-of-the-art fuzzy neural-based and traditional machine learning approaches. Results demonstrated that the proposed model is an efficient tool for these challenging classification tasks.
AB - Traditionally, fuzzy neural networks have parametric clustering methods based on equally spaced membership functions to fuzzify inputs of the model. In this sense, it produces an excessive number calculations for the parameters’ definition of the network architecture, which may be a problem especially for real-time large-scale tasks. Therefore, this paper proposes a new model that uses a non-parametric technique for the fuzzification process. The proposed model uses an autonomous data density approach in a pruned fuzzy neural network, wich favours the compactness of the model. The performance of the proposed approach is evaluated through the usage of databases related to the Optical Interconnection Network. Finally, binary patterns classification tests for the identification of temporal distribution (asynchronous or client–server) were performed and compared with state-of-the-art fuzzy neural-based and traditional machine learning approaches. Results demonstrated that the proposed model is an efficient tool for these challenging classification tasks.
KW - Autonomous data density
KW - Fuzzy neural networks
KW - Optical interconnection network
U2 - 10.1007/s12530-020-09336-3
DO - 10.1007/s12530-020-09336-3
M3 - Journal article
AN - SCOPUS:85084204599
VL - 12
SP - 899
EP - 911
JO - Evolving Systems
JF - Evolving Systems
SN - 1868-6478
IS - 4
ER -