PeleNet: A Reservoir Computing Framework for Loihi

Michaelis, Carlo

doi:10.48550/arxiv.2011.12338

Cited by 3 publications

(3 citation statements)

References 20 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Simulations were conducted with a custom software framework written in Python. The anisotropic network model code was based on an implementation by Leo Hiselius 82 in Brian2 83 with extended simulation management functionality inspired by the architecture from PeleNet 84 .…”

Section: Methodsmentioning

confidence: 99%

A model for cortical activity sequences

Lehr,

Erzmann,

Michaelis

et al. 2024

Preprint

Self Cite

View full text Add to dashboard Cite

Networks of neurons in the brain, that act on a timescale of milliseconds, can intrinsically generate reliable sequential activity on slow behavioral timescales of seconds. A possible mechanism for intrinsic sequence generation based on theoretical evidence points to distance-dependent connectivity with correlated spatial asymmetries, establishing an anisotropic network connectivity. We show that networks with such correlated asymmetric connectivity as well as symmetric distance-dependent connectivity match experimental data of connectivity motifs as well as neuronal activity statistics from rat and monkey cortex. At the network level, however, only the correlated asymmetric connectivity pattern generates spatiotemporal activity sequences on behaviorally relevant timescales, while the symmetric connectivity results in transient but stationary spatial bumps of neural activity. Our results strongly support the role of correlated asymmetries in connectivity for the generation of sequential activity in neural networks.

show abstract

Section: Methodsmentioning

confidence: 99%

A model for cortical activity sequences

Lehr,

Erzmann,

Michaelis

et al. 2024

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…At this point, without much background knowledge of neuromorphic hardware, one can get started programming using the various software development kits available (e.g., Brüderle et al, 2011 ; Sawada et al, 2016 ; Lin et al, 2018 ; Rhodes et al, 2018 ; Michaelis, 2020 ; Müller et al, 2020a , b ; Spilger et al, 2020 ; Rueckauer et al, 2021 ). Emulators for neuromorphic hardware (Furber et al, 2014 ; Petrovici et al, 2014 ; Luo et al, 2018 ; Valancius et al, 2020 ) running on a standard computer or field programmable gate arrays (FPGA), make it possible to develop neuromorphic network architectures without even needing access to a neuromorphic chip (see e.g., NengoLoihi 1 and Dynap-SE 2 ).…”

Section: Introductionmentioning

confidence: 99%

Brian2Loihi: An emulator for the neuromorphic chip Loihi using the spiking neural network simulator Brian

Michaelis

Lehr

Oed

et al. 2022

Front. Neuroinform.

Self Cite

View full text Add to dashboard Cite

Developing intelligent neuromorphic solutions remains a challenging endeavor. It requires a solid conceptual understanding of the hardware's fundamental building blocks. Beyond this, accessible and user-friendly prototyping is crucial to speed up the design pipeline. We developed an open source Loihi emulator based on the neural network simulator Brian that can easily be incorporated into existing simulation workflows. We demonstrate errorless Loihi emulation in software for a single neuron and for a recurrently connected spiking neural network. On-chip learning is also reviewed and implemented, with reasonable discrepancy due to stochastic rounding. This work provides a coherent presentation of Loihi's computational unit and introduces a new, easy-to-use Loihi prototyping package with the aim to help streamline conceptualization and deployment of new algorithms.

show abstract

Section: Introductionmentioning

confidence: 99%

Brian2Loihi: An emulator for the neuromorphic chip Loihi using the spiking neural network simulator Brian

Michaelis¹,

Lehr²,

Oed³

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

Developing intelligent neuromorphic solutions remains a challenging endeavour. It requires a solid conceptual understanding of the hardware's fundamental building blocks. Beyond this, accessible and user-friendly prototyping is crucial to speed up the design pipeline. We developed an open source Loihi emulator based on the neural network simulator Brian that can easily be incorporated into existing simulation workflows. We demonstrate errorless Loihi emulation in software for a single neuron and for a recurrently connected spiking neural network. On-chip learning is also reviewed and implemented, with reasonable discrepancy due to stochastic rounding. This work provides a coherent presentation of Loihi's computational unit and introduces a new, easy-to-use Loihi prototyping package with the aim to help streamline conceptualisation and deployment of new algorithms.

show abstract

PeleNet: A Reservoir Computing Framework for Loihi

Cited by 3 publications

References 20 publications

A model for cortical activity sequences

A model for cortical activity sequences

Brian2Loihi: An emulator for the neuromorphic chip Loihi using the spiking neural network simulator Brian

Brian2Loihi: An emulator for the neuromorphic chip Loihi using the spiking neural network simulator Brian

Contact Info

Product

Resources

About