Selection of Approximated Activation Functions in Neural Network-Based Sound Classifiers for Digital Hearing Aids

Álvarez L, Llerena C, Alexandre E, Gil-Pita R, Rosa-Zurera M. Selection of Approximated Activation Functions in Neural Network-Based Sound Classifiers for Digital Hearing Aids. In: AES Convention 130. Audio Engineering Society; 2011. Paper 8453. Available from: https://aes.org/publications/elibrary-page/?id=15920

@inproceedings{Alvarez2011_15920,
  author = {Álvarez, Lorena and Llerena, Cosme and Alexandre, Enrique and Gil-Pita, Roberto and Rosa-Zurera, Manuel},
  title = {{Selection of Approximated Activation Functions in Neural Network-Based Sound Classifiers for Digital Hearing Aids}},
  booktitle = {AES Convention 130},
  note = {Paper 8453},
  year = {2011},
  month = may,
  publisher = {Audio Engineering Society},
  url = {https://aes.org/publications/elibrary-page/?id=15920}
}

TY  - CPAPER
TI  - Selection of Approximated Activation Functions in Neural Network-Based Sound Classifiers for Digital Hearing Aids
AU  - Álvarez, Lorena
AU  - Llerena, Cosme
AU  - Alexandre, Enrique
AU  - Gil-Pita, Roberto
AU  - Rosa-Zurera, Manuel
T2  - AES Convention 130
M1  - Paper 8453
PY  - 2011
DA  - 2011/05/06
UR  - https://aes.org/publications/elibrary-page/?id=15920
PB  - Audio Engineering Society
LA  - en
AB  - The feasible implementation of signal processing techniques on hearing aids is constrained to the limited number of instructions per second to implement the algorithms on the digital signal processor the hearing aid is based on. This adversely limits the design of a neural network-based classifier embedded in the hearing aid. Aiming at helping the processor achieve accurate enough results, and in the effort of reducing the number of instructions per second, this paper focuses on exploring the most adequate approximations for the activation function. The experimental work proves that the approximated neural network-based classifier achieves the same efficiency as that reached by exact networks (without these approximations), but, this is the crucial point, with the added advantage of extremely reducing the computational cost on digital signal processor.
ER  -