Abstract
Activity-dependent plasticity has attracted the interest of researchers for years in the domain of computational neuroscience, as the modification of synaptic efficacy occurs as result of complex biochemical mechanisms that take place
at a cellular level. In this paper, we introduce a phenomenological model -implemented as an unsupervised learning rule for spiking neural networks- based on the cross-talk between glutamatergic and GABAergic neuroreceptors: NMDA, AMPA, GABAA, and GABAB. The proposed neuroreceptor-dependent
plasticity (NRDP) model is implemented and demonstrated in a spiking neural network environment, NeuCube, for modelling electroencephalography data. We show that the NRDP model can reproduce the generic spike-timing dependent plasticity behaviour in a spiking neural network. In addition, this can be used to simulate changes in excitatory/inhibitory balance in a spiking neural network by altering neuroreceptors activity. More specifically, by varying the parameters that affect neuroreceptors activation, we can study how these changes would affect the learning and memory ability of a subject. In a therapeutic context, this makes it a promising tool for studying the regulatory mechanisms where neuroreceptors cross-talk plays a crucial role. This can lead to new ways of early detection of neurological disorders and for better targeting drug treatments.
at a cellular level. In this paper, we introduce a phenomenological model -implemented as an unsupervised learning rule for spiking neural networks- based on the cross-talk between glutamatergic and GABAergic neuroreceptors: NMDA, AMPA, GABAA, and GABAB. The proposed neuroreceptor-dependent
plasticity (NRDP) model is implemented and demonstrated in a spiking neural network environment, NeuCube, for modelling electroencephalography data. We show that the NRDP model can reproduce the generic spike-timing dependent plasticity behaviour in a spiking neural network. In addition, this can be used to simulate changes in excitatory/inhibitory balance in a spiking neural network by altering neuroreceptors activity. More specifically, by varying the parameters that affect neuroreceptors activation, we can study how these changes would affect the learning and memory ability of a subject. In a therapeutic context, this makes it a promising tool for studying the regulatory mechanisms where neuroreceptors cross-talk plays a crucial role. This can lead to new ways of early detection of neurological disorders and for better targeting drug treatments.
Original language | English |
---|---|
Pages (from-to) | 63-72 |
Number of pages | 10 |
Journal | IEEE Transactions on Cognitive and Developmental Systems |
Volume | 11 |
Issue number | 1 |
Early online date | 22 Nov 2017 |
DOIs | |
Publication status | Published (in print/issue) - 1 Mar 2019 |
Keywords
- Artificial neural networks
- spiking neural networks
- unsupervised learning
- synaptic plasticity
- activity- dependent plasticity
- Hebbian rules
- neuroreceptors
- EEG data
- neurological disorders