On sensitivity of neural adaptive filters with respect to the slope parameter of a neuron activation function

Warren Sherliker, Igor R. Krcmar, Milorad Bozic, Danilo P. MandicIEEE International Conference on Acoustics Speech and Signal Processing 2002 · DOI

Abstract

Sensitivity analysis of neural adaptive filters with respect to the slope parameter of a neuron activation function is performed. The analysis is provided both for a feedforward neural adaptive filter and a recurrent perceptron. The slope affects stability and convergence characteristics of a filter via inherent relationship between the slope and the learning rate parameter. In addition, it determines character of an activation function, i.e. whether it is contractive or expansive mapping. Presented analysis shows that gradient-descent based learning algorithms with an adaptive learning rate significantly reduce sensitivity of a neural adaptive filter with respect to the slope parameter, when compared with learning algorithms with a constant learning rate. Experimental results on the test speech and HRV signals support the analysis.
BibTeX
@inproceedings{sherliker2002sensitivity,
  author = {Warren Sherliker and Igor R. Krcmar and Milorad Bozic and Danilo P. Mandic},
  title = {{On sensitivity of neural adaptive filters with respect to the slope parameter of a neuron activation function}},
  booktitle = {IEEE International Conference on Acoustics Speech and Signal Processing},
  year = {2002},
  doi = {10.1109/icassp.2002.5743978},
}

Loading related content...

Figure