Neural spiking dynamics in asynchronous digital circuits

Imam, Nabil; Wecker, Kyle; Tse, Jonathan; Karmazin, Robert; Manohar, Rajit

doi:10.1109/ijcnn.2013.6706952

Cited by 12 publications

(3 citation statements)

References 32 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…On the other hand, digital designs release the constraints on design time and sensitivity to noise, mismatch and PVT variations at the expense of going for a simulation approach lying further from the biophysics, thus inducing overall a large area penalty compared to analog designs. This is illustrated in the neuron implementation from [135] that implements a timestepped solver for the differential equations of the Izhikevich neuron model, while the phenomenological approach is followed in [136] with a 10-bit LIF neuron. Between both approaches lies the neuron model of Cassidy et al [137], it is based on a LIF neuron model to which configurability and stochasticity are added.…”

Section: ) Neurons (Soma)mentioning

confidence: 99%

Bottom-Up and Top-Down Neural Processing Systems Design: Neuromorphic Intelligence as the Convergence of Natural and Artificial Intelligence

Frenkel¹,

Bol²,

Indiveri³

2021

Preprint

View full text Add to dashboard Cite

While Moore's law has driven exponential computing power expectations, its nearing end calls for new avenues for improving the overall system performance. One of these avenues is the exploration of new alternative brain-inspired computing architectures that promise to achieve the flexibility and computational efficiency of biological neural processing systems. Within this context, neuromorphic intelligence represents a paradigm shift in computing based on the implementation of spiking neural network architectures tightly co-locating processing and memory. In this paper, we provide a comprehensive overview of the field, highlighting the different levels of granularity present in existing silicon implementations, comparing approaches that aim at replicating natural intelligence (bottom-up) versus those that aim at solving practical artificial intelligence applications (top-down), and assessing the benefits of the different circuit design styles used to achieve these goals. First, we present the analog, mixed-signal and digital circuit design styles, identifying the boundary between processing and memory through time multiplexing, in-memory computation and novel devices. Next, we highlight the key tradeoffs for each of the bottom-up and top-down approaches, survey their silicon implementations, and carry out detailed comparative analyses to extract design guidelines. Finally, we identify both necessary synergies and missing elements required to achieve a competitive advantage for neuromorphic edge computing over conventional machine-learning accelerators, and outline the key elements for a framework toward neuromorphic intelligence.

show abstract

Section: ) Neurons (Soma)mentioning

confidence: 99%

Bottom-Up and Top-Down Neural Processing Systems Design: Neuromorphic Intelligence as the Convergence of Natural and Artificial Intelligence

Frenkel¹,

Bol²,

Indiveri³

2021

Preprint

View full text Add to dashboard Cite

show abstract

“…For example, the natural communication in biological systems tends to be asynchronous and event driven. Therefore, an asynchronous communication protocol [43] coupled with an event driven approach [44] may potentially make the system more power efficient. Furthermore, sharing the common computing-path [45] (e.g., ALU1) and optimization of the neural network modularity [46], [47] will result in utilizing less hardware resources.…”

Section: Future Workmentioning

confidence: 99%

Optogenetics in Silicon: A Neural Processor for Predicting Optically Active Neural Networks

Luo

Nikolić

Evans

et al. 2017

IEEE Trans. Biomed. Circuits Syst.

View full text Add to dashboard Cite

We present a reconfigurable neural processor for real-time simulation and prediction of opto-neural behaviour. We combined a detailed Hodgkin-Huxley CA3 neuron integrated with a four-state Channelrhodopsin-2 (ChR2) model into reconfigurable silicon hardware. Our architecture consists of a Field Programmable Gated Array (FPGA) with a custom-built computing data-path, a separate data management system and a memory approach based router. Advancements over previous work include the incorporation of short and long-term calcium and light-dependent ion channels in reconfigurable hardware. Also, the developed processor is computationally efficient, requiring only 0.03 ms processing time per sub-frame for a single neuron and 9.7 ms for a fully connected network of 500 neurons with a given FPGA frequency of 56.7 MHz. It can therefore be utilized for exploration of closed loop processing and tuning of biologically realistic optogenetic circuitry.

show abstract

“…The literature is very diverse and growing fast, so we shall only cite a few examples here and refer the readers to further references in those works and in the review of Indiveri et al (2011): the digital processor chips TrueNorth developed by IBM (Cassidy et al, 2013;Merolla et al, 2014) and the more recent ODIN by ICTEAM (Frenkel et al, 2019); the compact neuron circuit, with only 14 MOSFET transistors proposed by Wijekoon and Dudek (2008); or the radically different spiking neuron based on vanadium dioxide (Yi et al, 2018), a Mott insulator memristive material (del Valle et al, 2018, del Valle et al, 2019. Other interesting proposals, which aimed at a faithful physical implementation of the Izhikevich mathematical model equations are: a compact circuit of MOS transistors in the subthreshold regime, simulated with MOSIS libraries (Rangan et al, 2010); a CMOS digital neuron for eventdriven computation, simulated in Spice (Imam et al, 2010).…”

Section: Introductionmentioning

confidence: 99%

Biologically Relevant Dynamical Behaviors Realized in an Ultra-Compact Neuron Model

2020

View full text Add to dashboard Cite

We demonstrate a variety of biologically relevant dynamical behaviors building on a recently introduced ultra-compact neuron (UCN) model. We provide the detailed circuits which all share a common basic block that realizes the leaky-integrate-andfire (LIF) spiking behavior. All circuits have a small number of active components and the basic block has only three, two transistors and a silicon controlled rectifier (SCR). We also demonstrate that numerical simulations can faithfully represent the variety of spiking behavior and can be used for further exploration of dynamical behaviors. Taking Izhikevich's set of biologically relevant behaviors as a reference, our work demonstrates that a circuit of a LIF neuron model can be used as a basis to implement a large variety of relevant spiking patterns. These behaviors may be useful to construct neural networks that can capture complex brain dynamics or may also be useful for artificial intelligence applications. Our UCN model can therefore be considered the electronic circuit counterpart of Izhikevich's (2003) mathematical neuron model, sharing its two seemingly contradicting features, extreme simplicity and rich dynamical behavior.

show abstract

Neural spiking dynamics in asynchronous digital circuits

Cited by 12 publications

References 32 publications

Bottom-Up and Top-Down Neural Processing Systems Design: Neuromorphic Intelligence as the Convergence of Natural and Artificial Intelligence

Bottom-Up and Top-Down Neural Processing Systems Design: Neuromorphic Intelligence as the Convergence of Natural and Artificial Intelligence

Optogenetics in Silicon: A Neural Processor for Predicting Optically Active Neural Networks

Biologically Relevant Dynamical Behaviors Realized in an Ultra-Compact Neuron Model

Contact Info

Product

Resources

About