Abstract
Over the past years Spiking Neural Networks (SNNs) models became attractive as a possible bridge to enable low-power event-driven neuromorphic hardware. SNNs have a high computational power due to the implicit employment of the biologically inspired input times. SNNs employ various parameters such as neuron threshold, synaptic delays, and weights in their structures. However, SNNs applications are still limited and elementary compared with other neural network architectures such as the Convolution Neural Networks (CNNs). In this research, a new SNN-based model named Adaptive Threshold Module (ATM) and its algorithm are proposed. The proposed ATM and algorithm depend on the adaptation of the internal spiking neuron threshold level. Adapting the threshold of the neurons is employed to control the spiking neuron firing rate to uniquely extract the main features of the input pattern that is in the shape of spike trains. It is shown that this technique works as an automated feature extraction method of input patterns in an efficient and faster way than other methods. The proposed method can preserve all information of the input spike trains. Simulations of the proposed model and the algorithm, using the challenging speech TIDIGITS dataset, sound RWCP dataset, and Poisson distribution spike trains, show encouraging results. The ATM can make SNN provide an accuracy surpassing that of the current state-of-the-art SNN algorithms and conventional non-spiking learning models.
Highlights
F OR more than half of a century, numerous models and structures of Artificial Neural Networks (ANNs) were considered as the cores of data processing automation in the fields of artificial intelligence and machine learning
ANNs passed to remarkable progress in the recent years after the milestone advances in the Deep Neural Networks (DNNs) with their learning algorithms [1], [2]
The Adaptive Threshold Module (ATM) is proposed to extract the main features of input patterns before being clustered by the classifier module
Summary
F OR more than half of a century, numerous models and structures of Artificial Neural Networks (ANNs) were considered as the cores of data processing automation in the fields of artificial intelligence and machine learning. All ANNs models are computational models that emulate the real biological neural networks in the brains of different living organisms. In the last two decades, more biological and realistic models of ANNs called Spiking Neural Networks (SNNs) emerged powerfully in research [3]. By adapting the threshold levels of the neurons, the output firing frequency is controlled This leads to the extraction of the input pattern features and information. The real-world input patterns are preprocessed and introduced into the ATM module in the shape of spike trains in the time domain Simulations show that this model works as a feature extraction of the input pattern, in a way almost similar to the CNN convolution and pooling layers and with fewer computations.
Sign-in/Register to view highlights & summary 
Published Version (
Free)
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call