Spatiotemporal Dynamics in Spiking Recurrent Neural Networks Using Modified-Full-FORCE on EEG Signals

Abstract

Methods on modelling the human brain as a Complex System have increased remarkably in the literature as researchers seek to understand the underlying foundations behind cognition, behaviour, and perception. Computational methods, especially Graph Theory-based methods, have recently contributed significantly in understanding wiring connectivity of the brain, modelling it as a set of nodes connected by edges. Therefore, the brain's spatiotemporal dynamics can be holistically studied by considering a network, which consists of many neurons, represented by nodes. Various models have been proposed for modelling such neurons. A recently proposed method in training such networks, called full-Force, produces networks that perform tasks with fewer neurons and greater noise robustness than previous least-squares approaches (i.e. FORCE method). In this paper, the first direct applicability of a variant of the full-Force method to biologically-motivated Spiking RNNs (SRNNs) is demonstrated. The SRNN is a graph consisting of modules. Each module is modelled as a Small-World Network (SWN), which is a specific type of a biologically-plausible graph. So, the first direct applicability of a variant of the full-Force method to modular SWNs is demonstrated, evaluated through regression and information theoretic metrics. For the first time, the aforementioned method is applied to spiking neuron models and trained on various real-life Electroencephalography (EEG) signals. To the best of the authors' knowledge, all the contributions of this paper are novel. Results show that trained SRNNs match EEG signals almost perfectly, while network dynamics can mimic the target dynamics. This demonstrates that the holistic setup of the network model and the neuron model which are both more biologically plausible than previous work, can be tuned into real biological signal dynamics.

Bayesian Deep Multi-Agent Multimodal Reinforcement Learning for Embedded Systems in Games, Natural Language Processing and Robotics

“…25) needed in order to bridge the gap towards the ultimate purpose of AI; however, these rules need to made 'learnable' (see Section 5.1). Hierarchical Modular (HM) models at the model-level are also found in the Computational Neuroscience literature [201,218], as shown in Fig. 24, thus, providing directions for drawing further inspiration from nature.…”

“…24, thus, providing directions for drawing further inspiration from nature. Blue dots show varying rewiring probability and impact on the Global efficiency (middle) and the Local efficiency (bottom) [201].…”

“…Drawing further inspiration from nature [201], in GUT-AI, the "black box" AI system should somehow encompass a small percentage (let's say between 10% to 20%) of 'oscillatory' determinism (through Dynamical Systems, Chaos Theory, Automata Theory, Category Theory, Topological Manifolds) instead of being purely stochastic, as it is the case with Deep…”

“…35). These rule-based 'constraints' (or rule-based 'priors') should not be confused with approaches in 'Physics-guided' or 'Physics-informed' ML, since the sole purpose of deterministic (or chaotic) rules is merely to enforce some kind of Dynamical Complexity [202], which tries to imitate the consciousness of the human brain in a bio-inspired way [201]. In other words, these rules have to be 'learnable' and not imposed by the human.…”

“…Figure 24 -Adjacency matrices depicting network organization for Hierarchical Modular Network andModular Networks (top)[201]. Adjacency matrices for four real-world weighted brain networks and one binary brain network (bottom)[218].…”