Versatile Emulation of Spiking Neural Networks on an Accelerated Neuromorphic Substrate

Billaudelle, Sebastian; Stradmann, Yannik; Schreiber, Korbinian; Cramer, Benjamin; Baumbach, Andreas; Dold, Dominik; Göltz, Julian; Kungl, Ákos F.; Wunderlich, Timo C.; Hartel, Andreas; Müller, Eric; Breitwieser, Oliver; Mauch, Christian; Kleider, Mitja; Grübl, Andreas; Stöckel, David; Pehle, Christian; Heimbrecht, Arthur; Spilger, Philipp; Kiene, Gerd; Karasenko, Vitali; Senn, Walter; Petrovici, Mihai A.; Schemmel, Johannes; Meier, K.

doi:10.1109/iscas45731.2020.9180741

Cited by 34 publications

(26 citation statements)

References 36 publications

(39 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Missing pixels correspond to locations not reached at any time. (A–E) were adapted from Schreiber ( 2021 ) and (C) from Billaudelle et al ( 2020 ).…”

Section: Applications Of the Brainscales-2 Systemmentioning

confidence: 99%

The BrainScaleS-2 Accelerated Neuromorphic System With Hybrid Plasticity

et al. 2022

View full text Add to dashboard Cite

Since the beginning of information processing by electronic components, the nervous system has served as a metaphor for the organization of computational primitives. Brain-inspired computing today encompasses a class of approaches ranging from using novel nano-devices for computation to research into large-scale neuromorphic architectures, such as TrueNorth, SpiNNaker, BrainScaleS, Tianjic, and Loihi. While implementation details differ, spiking neural networks—sometimes referred to as the third generation of neural networks—are the common abstraction used to model computation with such systems. Here we describe the second generation of the BrainScaleS neuromorphic architecture, emphasizing applications enabled by this architecture. It combines a custom analog accelerator core supporting the accelerated physical emulation of bio-inspired spiking neural network primitives with a tightly coupled digital processor and a digital event-routing network.

show abstract

“…Missing pixels correspond to locations not reached at any time. (A–E) were adapted from Schreiber ( 2021 ) and (C) from Billaudelle et al ( 2020 ).…”

Section: Applications Of the Brainscales-2 Systemmentioning

confidence: 99%

The BrainScaleS-2 Accelerated Neuromorphic System With Hybrid Plasticity

et al. 2022

View full text Add to dashboard Cite

show abstract

“…Spatial noise reflects the individuality of cortical neurons or the heterogeneity arising from device mismatch in hardware. Here, we focus on the heterogeneity of time constants; in contrast to, for example, variability in synaptic parameters or activation functions, these variations can not be "trained away" by adapting synaptic weights [36][37][38]. The two time constants that govern neuron dynamics in our model, namely integration (Eqn.…”

Section: Robustness To Substrate Imperfectionsmentioning

confidence: 99%

Latent Equilibrium: A unified learning theory for arbitrarily fast computation with arbitrarily slow neurons

Haider¹,

Ellenberger²,

Kriener³

et al. 2021

Preprint

View full text Add to dashboard Cite

The response time of physical computational elements is finite, and neurons are no exception. In hierarchical models of cortical networks each layer thus introduces a response lag. This inherent property of physical dynamical systems results in delayed processing of stimuli and causes a timing mismatch between network output and instructive signals, thus afflicting not only inference, but also learning. We introduce Latent Equilibrium, a new framework for inference and learning in networks of slow components which avoids these issues by harnessing the ability of biological neurons to phase-advance their output with respect to their membrane potential. This principle enables quasi-instantaneous inference independent of network depth and avoids the need for phased plasticity or computationally expensive network relaxation phases. We jointly derive disentangled neuron and synapse dynamics from a prospective energy function that depends on a network's generalized position and momentum. The resulting model can be interpreted as a biologically plausible approximation of error backpropagation in deep cortical networks with continuous-time, leaky neuronal dynamics and continuously active, local plasticity. We demonstrate successful learning of standard benchmark datasets, achieving competitive performance using both fully-connected and convolutional architectures, and show how our principle can be applied to detailed models of cortical microcircuitry. Furthermore, we study the robustness of our model to spatio-temporal substrate imperfections to demonstrate its feasibility for physical realization, be it in vivo or in silico.

show abstract

“…Such single-spike coding enables fast information processing by explicitly encouraging the emission of as few spikes as early as possible, which meets physiological constraints and reaction times observed in humans and animals [42][43][44][45]. Apart from biological plausibility, such a fast and sparse coding scheme is a natural fit for neuromorphic systems that offer energy-efficient and fast emulation of spiking neural networks [46][47][48][49][50][51][52].…”

Section: Introductionmentioning

confidence: 99%

“…As our algorithm only requires knowledge about afferent and efferent spike times of all neurons, it lends itself to emulation on neuromorphic hardware. The accelerated, yet power-efficient BrainScaleS-2 platform [48,59] pairs especially well with the sparseness and low latency already inherent to TTFS coding. We show how an implementation of our algorithm on BrainScaleS-2 can obtain similar classification accuracies to software simulations, while displaying highly competitive time and power characteristics, with a combination of 48 µs and 8.4 µJ per classification.…”

Section: Introductionmentioning

confidence: 99%

Fast and energy-efficient neuromorphic deep learning with first-spike times

et al. 2021

Self Cite

View full text Add to dashboard Cite

For a biological agent operating under environmental pressure, energy consumption and reaction times are of critical importance. Similarly, engineered systems are optimized for short time-to-solution and low energy-to-solution characteristics. At the level of neuronal implementation, this implies achieving the desired results with as few and as early spikes as possible. With time-to-first-spike coding both of these goals are inherently emerging features of learning.Here, we describe a rigorous derivation of a learning rule for such first-spike times in networks of leaky integrate-andfire neurons, relying solely on input and output spike times, and show how this mechanism can implement error backpropagation in hierarchical spiking networks. Furthermore, we emulate our framework on the BrainScaleS-2 neuromorphic system and demonstrate its capability of harnessing the system's speed and energy characteristics. Finally, we examine how our approach generalizes to other neuromorphic platforms by studying how its performance is affected by typical distortive effects induced by neuromorphic substrates.

show abstract

Versatile Emulation of Spiking Neural Networks on an Accelerated Neuromorphic Substrate

Cited by 34 publications

References 36 publications

The BrainScaleS-2 Accelerated Neuromorphic System With Hybrid Plasticity

The BrainScaleS-2 Accelerated Neuromorphic System With Hybrid Plasticity

Latent Equilibrium: A unified learning theory for arbitrarily fast computation with arbitrarily slow neurons

Fast and energy-efficient neuromorphic deep learning with first-spike times

Contact Info

Product

Resources

About