Simplified spiking neural network architecture and STDP learning algorithm applied to image classification

Abstract

Spiking neural networks (SNN) have gained popularity in embedded applications such as robotics and computer vision. The main advantages of SNN are the temporal plasticity, ease of use in neural interface circuits and reduced computation complexity. SNN have been successfully used for image classification. They provide a model for the mammalian visual cortex, image segmentation and pattern recognition. Different spiking neuron mathematical models exist, but their computational complexity makes them ill-suited for hardware implementation. In this paper, a novel, simplified and computationally efficient model of spike response model (SRM) neuron with spike-time dependent plasticity (STDP) learning is presented. Frequency spike coding based on receptive fields is used for data representation; images are encoded by the network and processed in a similar manner as the primary layers in visual cortex. The network output can be used as a primary feature extractor for further refined recognition or as a simple object classifier. Results show that the model can successfully learn and classify black and white images with added noise or partially obscured samples with up to ×20 computing speed-up at an equivalent classification ratio when compared to classic SRM neuron membrane models. The proposed solution combines spike encoding, network topology, neuron membrane model and STDP learning.

A computational model of loss of dopaminergic cells in Parkinson’s disease due to glutamate-induced excitotoxicity

“…In the future, we will incorporate a detailed biophysical model of SNc along with dopamine synthesis pathway (Tello-Bravo, 2012), apoptosis pathway (Hong et al, 2012) and neural energy supply-consumption properties to the current model. The synaptic weights in the proposed model are not dynamic, we would like to include some type of learning principle by incorporating STDP type of learning rule in STN population which can show the long-term effect of DBS treatment (Ebert et al, 2014;Iakymchuk et al, 2015).…”

Wide learning: Using an ensemble of biologically-plausible spiking neural networks for unsupervised parallel classification of spatio-temporal patterns

“…In the current form, the ensemble has a capacity to generalise and a limited capacity to process noisy, incomplete data, which has been previously observed in some SNN [29], [30]. However, in our experiments we used random noise and complete silencing of a proportion of input neurons, which are extreme cases of data corruption as there is no link between the corrupted values and the original data.…”

A Multi-Scale Computational Model of Excitotoxic Loss of Dopaminergic Cells in Parkinson’s Disease

“…In order to do so, all neuronal types should be modeled as conductance-based models where calcium dynamics should be considered. The synaptic weights in the proposed model are not dynamic; we would like to incorporate learning principles such as spike time-dependent plasticity in the STN population, which can show the long-term effect of DBS treatment (Ebert et al, 2014;Iakymchuk et al, 2015). A more ambitious goal is to incorporate the present model into a large-scale model of basal ganglia (Muralidharan et al, 2018) in understanding the effect of therapeutics in terms of the behavioral response (Erdi et al, 2006).…”