Title :
Feedforward sigmoidal networks - equicontinuity and fault-tolerance properties
Author :
Chandra, Pravin ; Singh, Yogesh
Author_Institution :
Sch. of Inf. Technol., GGS Indraprastha Univ., Delhi, India
Abstract :
Sigmoidal feedforward artificial neural networks (FFANNs) have been established to be universal approximators of continuous functions. The universal approximation results are summarized to identify the function sets represented by the sigmoidal FFANNs with the universal approximation properties. The equicontinuous properties of the identified sets is analyzed. The equicontinuous property is related to the fault tolerance of the sigmoidal FFANNs. The generally used arbitrary weight sigmoidal FFANNs are shown to be nonequicontinuous sets. A class of bounded weight sigmoidal FFANNs is established to be equicontinuous. The fault-tolerance behavior of the networks is analyzed and error bounds for the induced errors established.
Keywords :
fault tolerance; feedforward neural nets; function approximation; equicontinuity property; fault-tolerance property; sigmoidal feedforward artificial neural network; universal approximator; Artificial neural networks; Biological neural networks; Biological system modeling; Biology computing; Computer networks; Error analysis; Fault tolerance; Information technology; Space technology; Writing; Equicontinuity; fault-tolerance; feedforward artificial neural networks (FFANNs); function sets; sigmoidal networks; Algorithms; Artificial Intelligence; Computer Simulation; Decision Support Techniques; Feedback; Logistic Models; Neural Networks (Computer); Pattern Recognition, Automated; Reproducibility of Results; Sensitivity and Specificity;
Journal_Title :
Neural Networks, IEEE Transactions on
DOI :
10.1109/TNN.2004.831198
