Artificial neurovascular network (ANVN) to study the accuracy vs. efficiency trade-off in an energy dependent neural network

Abstract

Artificial feedforward neural networks perform a wide variety of classification and function approximation tasks with high accuracy. Unlike their artificial counterparts, biological neural networks require a supply of adequate energy delivered to single neurons by a network of cerebral microvessels. Since energy is a limited resource, a natural question is whether the cerebrovascular network is capable of ensuring maximum performance of the neural network while consuming minimum energy? Should the cerebrovascular network also be trained, along with the neural network, to achieve such an optimum? In order to answer the above questions in a simplified modeling setting, we constructed an Artificial Neurovascular Network (ANVN) comprising a multilayered perceptron (MLP) connected to a vascular tree structure. The root node of the vascular tree structure is connected to an energy source, and the terminal nodes of the vascular tree supply energy to the hidden neurons of the MLP. The energy delivered by the terminal vascular nodes to the hidden neurons determines the biases of the hidden neurons. The “weights” on the branches of the vascular tree depict the energy distribution from the parent node to the child nodes. The vascular weights are updated by a kind of “backpropagation” of the energy demand error generated by the hidden neurons. We observed that higher performance was achieved at lower energy levels when the vascular network was also trained along with the neural network. This indicates that the vascular network needs to be trained to ensure efficient neural performance. We observed that below a certain network size, the energetic dynamics of the network in the per capita energy consumption vs. classification accuracy space approaches a fixed-point attractor for various initial conditions. Once the number of hidden neurons increases beyond a threshold, the fixed point appears to vanish, giving place to a line of attractors. The model also showed that when there is a limited resource, the energy consumption of neurons is strongly correlated to their individual contribution to the network’s performance.

The development of bi‐directionally coupled self‐organizing neurovascular networks captures orientation‐selective neural and hemodynamic cortical responses

“…This assumption forms the basis of interpreting functional neuroimaging techniques (Raichle, 2000(Raichle, , 2009Raichle & Mintun, 2006). However, recent years have shown that vascular feedback may influence neural signals in critical ways (Chander & Chakravarthy, 2012;Chhabria & Chakravarthy, 2016;Kim et al, 2016;Kumar, Mayakkannan, et al, 2021;Ranasinghe et al, 2015;Roy et al, 2021). Here, we modelled interactions between neurons and blood vessels, including a lateral connectivity network among vessels, to shed light on how vascular dynamics might generate tuned vascular responses and how these responses affect neuronal network structures.…”

“…For instance, recent experimental observations support the idea that cerebral microvessels can influence information coding in neurons (Kim et al, 2016; Roy et al, 2021). Additionally, computational models developed by our group suggest that information processing in the cerebral microvasculature may occur in tandem with neural information processing (Chander & Chakravarthy, 2012; Chhabria & Chakravarthy, 2016; Kumar, Khot, et al, 2021; Kumar, Mayakkannan, et al, 2021). Therefore, we included a feedback component from the vasculature to the neural layer and sought to determine if this could affect neuronal properties.…”

The Influence of Neural Activity and Neural Cytoarchitecture on Cerebrovascular Arborization: A Computational Model

“…The relevance of a healthy vasculature in shaping the function of the neural network has been discussed extensively ( Moore and Cao, 2008 ; Pradhan and Chakravarthy, 2011 ; Chander and Chakravarthy, 2012 ; Chhabria and Chakravarthy, 2016 ; Tourigny et al, 2019 ; Bright et al, 2020 ; Kumar et al, 2021a , b ). Neurodegenerative diseases are being traced back to an impaired neurovascular functional coupling ( Iadecola, 2004 , 2017 ; Shabir et al, 2018 ; Ahmad et al, 2020 ; Muddapu et al, 2020 ).…”

The Influence of Neural Activity and Neural Cytoarchitecture on Cerebrovascular Arborization: A Computational Model

“…The relevance of a healthy vasculature in shaping the function of the neural network has been discussed extensively [39][40][41][42][43][44][45][46]. Neurodegenerative diseases are being traced back to an impaired neurovascular functional coupling [1,[47][48][49][50].…”