Empowering the impaired astrocytes in the tripartite synapses to improve accuracy of pattern recognition

Abstract

Recent finding has demonstrated that glial cells and especially astrocytes are responsible for some important roles in the central nervous system. Laboratory examination of the roles of astrocytes in information processing of neural system is associated with technical difficulties. Therefore, computational modeling provides a suitable approach in describing cognitive phenomena and neuroscience. In this paper, the role of strong and weak astrocytes in pattern recognition has considered and bio-stimulator was used to compensate the malfunction of the weak astrocyte. Therefore, we designed the tripartite synapses network, which astrocytes dominate the synaptic spaces. Then, the network has been trained on MNIST dataset. After that, the control feedback of astrocyte to pre- and postsynaptic neurons has decreased so that the astrocyte dominance on synaptic transitions decreased and impaired tripartite synapses network has been configured. In the next step, due to the destructive effects of impaired astrocyte, based on the dynamical model of the astrocyte biophysical model, a bio-inspired stimulator was designed. Accordingly, a new concept “Stimulating the impaired astrocytes to compensate their malfunction in pattern recognition” was introduced. This new mechanism was proposed to stimulate the impaired astrocytes in a population of tripartite synapses for restoration of the normal neural oscillations. Reported results confirmed that the increased error of “always firing” in pattern recognition caused by astrocytes dysfunction could be compensated by the bio-inspired stimulator.