Deploying and Optimizing Embodied Simulations of Large-Scale Spiking Neural Networks on HPC Infrastructure

Abstract

Simulating the brain-body-environment trinity in closed loop is an attractive proposal to investigate how perception, motor activity and interactions with the environment shape brain activity, and vice versa. The relevance of this embodied approach, however, hinges entirely on the modeled complexity of the various simulated phenomena. In this article, we introduce a software framework that is capable of simulating large-scale, biologically realistic networks of spiking neurons embodied in a biomechanically accurate musculoskeletal system that interacts with a physically realistic virtual environment. We deploy this framework on the high performance computing resources of the EBRAINS research infrastructure and we investigate the scaling performance by distributing computation across an increasing number of interconnected compute nodes. Our architecture is based on requested compute nodes as well as persistent virtual machines; this provides a high-performance simulation environment that is accessible to multi-domain users without expert knowledge, with a view to enable users to instantiate and control simulations at custom scale via a web-based graphical user interface. Our simulation environment, entirely open source, is based on the Neurorobotics Platform developed in the context of the Human Brain Project, and the NEST simulator. We characterize the capabilities of our parallelized architecture for large-scale embodied brain simulations through two benchmark experiments, by investigating the effects of scaling compute resources on performance defined in terms of experiment runtime, brain instantiation and simulation time. The first benchmark is based on a large-scale balanced network, while the second one is a multi-region embodied brain simulation consisting of more than a million neurons and a billion synapses. Both benchmarks clearly show how scaling compute resources improves the aforementioned performance metrics in a near-linear fashion. The second benchmark in particular is indicative of both the potential and limitations of a highly distributed simulation in terms of a trade-off between computation speed and resource cost. Our simulation architecture is being prepared to be accessible for everyone as an EBRAINS service, thereby offering a community-wide tool with a unique workflow that should provide momentum to the investigation of closed-loop embodiment within the computational neuroscience community.

Deep neural generation of neuronal spikes

“…It should also be emphasized that even if one creates a detailed computational model of neural systems, it is actually very difficult to prepare a computational model that generates spikes that somehow reproduce the synchrony among the many pairs of neurons in the system [Nolte et al, 2019; Dura-Bernal et al, 2019; Feldotto et al, 2022]. The findings of this study can be summarized in the following three categories:…”

A Spiking Neural Network Builder for Systematic Data-to-Model Workflow

“…Big spike data could be loaded into the database and meticulously queried and plotted for better interpretation of results. While small simulations could be triggered instantly, large simulations can be prepared and dispatched for high-performance computing, as shown in Feldotto et al ( 2022 ). SNNbuilder straightforward code generation may include job scripts for the setup and simulation of large-scale models on the Fugaku supercomputer (Sato et al, 2020 ), as well on collaborative simulation infrastructures like Fenix from EBRAINS (Alam et al, 2019 ).…”

UPDATE trial: investigating the effects of ultra-processed versus minimally processed diets following UK dietary guidance on health outcomes: a protocol for an 8-week community-based cross-over randomised controlled trial in people with overweight or obesity, followed by a 6-month behavioural intervention

“…However, the neurobiological impact of high-UPF diets is largely unknown. A subset of participants (n=24) will undergo an advanced MRI brain protocol at the baseline and 8-week visits of the RCT (four visits), to assess a number of changes in brain properties: (1) functional resting-state connectivity between regions implicated in eating behaviour, metabolism and swelling linked to diffuse inflammation, and their reversibility66; (2) microstructure brain properties extracted from diffusion weighted imaging data67; (3) changes in brain dynamics, based on each subject’s functional and structural connectivity68; (4) metabolite changes, such as glutamate, N-acetyl aspartate and Inositol, through brainstem MR spectroscopy69; (5) regional volume changes through high-resolution structural scans66 and (6) relaxometry properties alterations affected by inflammation 70…”