TY - JOUR
T1 - Spiking Neural Network Model of Sound Localisation using the interaural intensity Difference
AU - Wall, Julie
AU - McDaid, Liam
AU - Maguire, Liam
AU - McGinnity, TM
PY - 2012/4
Y1 - 2012/4
N2 - Abstract—In this paper, a spiking neural network (SNN) architecture to simulate the sound localisation ability of the mammalian auditory pathways using the interaural intensitydifference (IID) cue is presented. The lateral superior olive (LSO) was the inspiration for the architecture which required the integration of an auditory periphery (cochlea) model and a model of the medial nucleus of the trapezoid body (MNTB). The SNN uses leaky integrate and fire excitatory and inhibitory spiking neurons, facilitating synapses and receptive fields. Experimentally derived Head Related Transfer Function (HRTF) acoustical datafrom adult domestic cats were employed to train and validate the localisation ability of the architecture; training used the supervised learning algorithm called the Remote Supervision Method (ReSuMe) to determine the azimuthal angles. The experimental results demonstrate that the architecture performs best when it is localising high frequency sound data in agreement with the biology, and also shows a high degree of robustness when theHRTF acoustical data is corrupted by noise.Index Terms—Spiking neural networks, sound localisation, lateral superior olive, interaural intensity difference
AB - Abstract—In this paper, a spiking neural network (SNN) architecture to simulate the sound localisation ability of the mammalian auditory pathways using the interaural intensitydifference (IID) cue is presented. The lateral superior olive (LSO) was the inspiration for the architecture which required the integration of an auditory periphery (cochlea) model and a model of the medial nucleus of the trapezoid body (MNTB). The SNN uses leaky integrate and fire excitatory and inhibitory spiking neurons, facilitating synapses and receptive fields. Experimentally derived Head Related Transfer Function (HRTF) acoustical datafrom adult domestic cats were employed to train and validate the localisation ability of the architecture; training used the supervised learning algorithm called the Remote Supervision Method (ReSuMe) to determine the azimuthal angles. The experimental results demonstrate that the architecture performs best when it is localising high frequency sound data in agreement with the biology, and also shows a high degree of robustness when theHRTF acoustical data is corrupted by noise.Index Terms—Spiking neural networks, sound localisation, lateral superior olive, interaural intensity difference
U2 - 10.1109/TNNLS.2011.2178317
DO - 10.1109/TNNLS.2011.2178317
M3 - Article
VL - 23
SP - 574
EP - 586
JO - IEEE Transactions on Neural Networks and Learning Systems
JF - IEEE Transactions on Neural Networks and Learning Systems
IS - 4
ER -