Real-Time Simulation of a Cerebellar Scaffold Model on Graphics Processing Units

Abstract

Large-scale simulation of detailed computational models of neuronal microcircuits plays a prominent role in reproducing and predicting the dynamics of the microcircuits. To reconstruct a microcircuit, one must choose neuron and synapse models, placements, connectivity, and numerical simulation methods according to anatomical and physiological constraints. For reconstruction and refinement, it is useful to be able to replace one module easily while leaving the others as they are. One way to achieve this is via a scaffolding approach, in which a simulation code is built on independent modules for placements, connections, and network simulations. Owing to the modularity of functions, this approach enables researchers to improve the performance of the entire simulation by simply replacing a problematic module with an improved one. Casali et al. (2019) developed a spiking network model of the cerebellar microcircuit using this approach, and while it reproduces electrophysiological properties of cerebellar neurons, it takes too much computational time. Here, we followed this scaffolding approach and replaced the simulation module with an accelerated version on graphics processing units (GPUs). Our cerebellar scaffold model ran roughly 100 times faster than the original version. In fact, our model is able to run faster than real time, with good weak and strong scaling properties. To demonstrate an application of real-time simulation, we implemented synaptic plasticity mechanisms at parallel fiber–Purkinje cell synapses, and carried out simulation of behavioral experiments known as gain adaptation of optokinetic response. We confirmed that the computer simulation reproduced experimental findings while being completed in real time. Actually, a computer simulation for 2 s of the biological time completed within 750 ms. These results suggest that the scaffolding approach is a promising concept for gradual development and refactoring of simulation codes for large-scale elaborate microcircuits. Moreover, a real-time version of the cerebellar scaffold model, which is enabled by parallel computing technology owing to GPUs, may be useful for large-scale simulations and engineering applications that require real-time signal processing and motor control.

The quest for multiscale brain modeling

“…These models have been simplified, while maintaining the salient aspects of spike discharge (but not so far of dendritic processing; e.g., [130]), and have been used to generate SNNs [78]. Finally, these SNNs have been embedded into hybrid closed-loop controllers [50,[78][79][80]119,120] and transformed into neuromorphic hardware [81] that is capable of running very large-scale simulations [131]. As a whole, these controllers are providing remarkable support to cerebellar theory, physiology, and pathology [41,119,125].…”

“…It has been estimated that simulating a whole human brain at cellular resolution would require up to ~4 × 10 29 TFLOPS, whereas the Fugaku Supercomputer at RIKEN 8 (Japan) has a peak performance of ~5 × 10 5 TFLOPS [6]. One open perspective is to transform digital brain simulators in neuromorphic hardware to bring computation close to real time [79,81,131]. Overall, personalizing a brain model remains difficult, thus impacting on the challenging perspective of generating brain digital twins [22,133], as discussed in the following text.…”

A cerebellum inspired spiking neural network as a multi-model for pattern classification and robotic trajectory prediction

“…The current model was reconstructed based on known microcircuitry of the cerebellum ( Eccles, 1967 ), electrophysiological behavior ( D’Angelo et al, 2001 ), and significant plasticity rules ( Mapelli et al, 2015 ). Even though there are several models available that look at the different aspects of the cerebellum ( Casellato et al, 2012 ; Garrido et al, 2013 ; Antonietti et al, 2015 ; D’Angelo et al, 2016b ; Luque et al, 2016 ; Yamaura et al, 2020 ; Kuriyama et al, 2021 ), the present model covers certain aspects of the cerebellum while some loops are skipped. Initial models looked at only single-layered neurons ( Marr, 1969 ; Albus, 1971 ) which were extended with many other layers and plasticity rules.…”

“…Spiking neural networks (SNN) exploit a biologically observed phenomenological element in ML, allowing optimization and parallelizability to algorithms ( Naveros et al, 2017 ) which may be event-driven and time-driven and may incorporate spatiotemporal information processing capabilities of biological neural circuits. Algorithms that are based on different brain circuits, such as the visual cortex ( Fu et al, 2012 ; Yamins and DiCarlo, 2016 ), basal ganglia ( Doya, 2000 ; Baladron and Hamker, 2015 ; Girard et al, 2020 ), and cerebellum ( Casellato et al, 2012 ; Garrido et al, 2013 ; Antonietti et al, 2015 ; D’Angelo et al, 2016b ; Luque et al, 2016 ; Yamaura et al, 2020 ; Kuriyama et al, 2021 ) with spiking neural models help understand the circuitry and in turn, help reconstruct and train systems. EDLUT ( Ros et al, 2006 ), SpiNNaker ( Khan et al, 2008 ), MuSpiNN ( Ghosh-Dastidar and Adeli, 2009 ), and biCNN ( Pinzon-Morales and Hirata, 2013 ) are some of the existing brain-inspired models which are used in the field of control systems for robotic articulation control.…”

Artificial cerebellum on FPGA: Realistic real-time cerebellar spiking neural network model capable of real-world adaptive motor control

“…When the membrane potential 𝑣(𝑡) exceeds the threshold 𝑉 𝑡ℎ , it is reset to the resting potential 𝑉 𝑟 and the spike event function 𝑠(𝑡) outputs 1. 𝑖 spont (𝑡) is a uniform random number [0, 𝐼 spont ] to describe a dc firing rate of each neuron. The constants for each neuron/fiber type are listed in Table 1 which are the same as the previous realistic artificial cerebellums (Casali et al, 2019;Kuriyama et al, 2021) except that the parameters of the input fibers were arbitrarily defined so that their firing frequencies become physiologically appropriate.…”