A comparative study between E‐neurons mathematical model and circuit model

Daliri, Mojtaba; Ferreira, Pietro Maris; Klisnick, Geoffroy; Benlarbi-Delaï, Aziz

doi:10.1049/cds2.12017

Cited by 4 publications

(3 citation statements)

References 20 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…where V m is the spiking voltage signal; i ex is the input excitation current; V DD is the eNeuron supply voltage; η is the subthreshold slope; φ t is the thermal voltage (k b T /q). A detailed modeling is presented in [22]. Figure 2(a) illustrates a LIF eNeuron proposed by Danneville et al, which is an Axon-Hillock topology [7].…”

Section: B Analog Spiking Neural Network: Electronic Neuronsmentioning

confidence: 99%

Analog Spiking Neural Network Synthesis for the MNIST

Soupizet

Jouni

Wang³

et al. 2023

JICS

View full text Add to dashboard Cite

Different from classical artificial neural network which processes digital data, the spiking neural network (SNN) processes spike trains. Indeed, its event-driven property helps to capture the rich dynamics the neurons have within the brain, and the sparsity of collected spikes helps reducing computational power. Novel synthesis framework is proposed and an algorithm is detailed to guide designers into deep learning and energy-efficient analog SNN using MNIST. An analog SNN composed of 86 electronic neurons (eNeuron) and 1238 synapses interacting through two hidden layers is illustrated. Three different models of eNeurons implementations are tested, being (Leaky) Integrate-and-Fire (LIF), Morris Lecar (ML) simplified (simp.) and biomimetic (bio.). The proposed SNN, coupling deep learning and ultra-low power, is trained using a common machine learning system (Tensor- Flow) for the MNIST. LIF eNeurons implementations present some limitations and weakness in terms of dynamic range. Both ML eNeurons achieve robust accuracy which is approximately of 0.82.

show abstract

Section: B Analog Spiking Neural Network: Electronic Neuronsmentioning

confidence: 99%

Analog Spiking Neural Network Synthesis for the MNIST

Soupizet

Jouni

Wang³

et al. 2023

JICS

View full text Add to dashboard Cite

show abstract

Section: A Neuronsmentioning

confidence: 99%

Deep Neural Network Feasibility Using Analog Spiking Neurons

Soupizet

Jouni

Sulzbach

et al. 2022

2022 35th SBC/SBMicro/IEEE/ACM Symposium on Integrated Circuits and Systems Design (SBCCI)

Self Cite

View full text Add to dashboard Cite

Novel non-Von-Neumann solutions have raised based on artificial intelligence (AI) such as the neuromorphic spiking processors in either analog or digital domain. This paper proposes to study the deep neural network feasibility using ultra-low-power eNeuron. The trade-offs in terms of deep learning capabilities and energy efficiency are highlighted. A linear fit model is found in the region of high energy efficiency of neuromorphic components. Thus, deep learning and energy efficiency mutually exclusive if those neuromorphic components are used.

show abstract

“…On the other hand, in the biological neuron, the peculiarity lies in its autonomy in generating the output. The I&F neuron circuit realizes the autonomous pulse generation in the hardware manner in a way that it can fire an output signal modulating the interneuron connectivity only when the threshold is reached as the accumulation of charges is progressed [14], [15], [16]. In this sense, stability and robustness of a neuron circuit against the variation in electrical signals from the synapse array are essential for better predictability and reliability of the system.…”

Section: Introductionmentioning

confidence: 99%

A Compact Integrate-and-Fire Neuron Circuit Embedding Operational Transconductance Amplifier for Fidelity Enhancement

et al. 2023

View full text Add to dashboard Cite

In this study, a compact CMOS integrate-and-fire (I&F) neuron circuit embedding an operational transconductance amplifier (OTA) has been designed for enhancing the fidelity in output generation. The OTA block in the neuron circuit allows for maintaining stability in I&F functions even under high-frequency operation conditions. The designed neuron circuit consists of OTA circuit, membrane capacitor, inverter, and reset MOSFET, from which the area occupancy is approximated to be 22 × 43 µm 2 . Featuring the simple and compact structure, the proposed neuron circuit shows the capability to control the firing frequency by adjusting the amplitude and temporal width of the synaptic pulse, resulting in high fidelity in I&F function. Series of circuit simulations have been performed to validate the systematic operations of the neuron circuit by HSPICE presuming the 0.35-µm Si CMOS technology. Moreover, temperature dependence was also investigated so that the robustness and stability of the neuron circuit at elevated operation temperatures were verified. The results provide a practical way of designing a compact and reliable neuron circuit working with the synaptic devices having deviations in operation characteristics in the hardware-oriented spiking neural network (SNN).INDEX TERMS Integrate-and-fire neuron circuit, operational transconductance amplifier (OTA), fidelity, circuit simulation, stability, hardware-oriented spiking neural network (SNN).

show abstract

A comparative study between E‐neurons mathematical model and circuit model

Cited by 4 publications

References 20 publications

Analog Spiking Neural Network Synthesis for the MNIST

Analog Spiking Neural Network Synthesis for the MNIST

Deep Neural Network Feasibility Using Analog Spiking Neurons

A Compact Integrate-and-Fire Neuron Circuit Embedding Operational Transconductance Amplifier for Fidelity Enhancement

Contact Info

Product

Resources

About