Pith. sign in

REVIEW 4 major objections 7 minor 61 references

An open-source framework embeds real floating-gate and ReRAM device physics into SNN training so designers can optimize physical synaptic parameters and jointly score accuracy, area and power.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

T0 review · grok-4.5

2026-07-14 16:00 UTC pith:6DVK6ISY

load-bearing objection Useful open mixed-signal SNN co-design tool with real device-parameter training; the hardware-predictive claim is only weakly anchored beyond single-neuron ISI and a tiny XOR net. the 4 major comments →

arxiv 2607.06456 v2 pith:6DVK6ISY submitted 2026-07-07 eess.SP cs.NE

A Hardware-Aware Open-Source Framework for Design Space Exploration of Mixed-Signal Spiking Neural Networks

Sayma Nowshin Chowdhury , Vineeta Nair , Taseen Forhad , Aishwarya Natarajan , Corey Hart , Sahil Shah This is my paper
classification eess.SP cs.NE
keywords mixed-signal SNNshardware-aware trainingfloating-gate synapsesReRAMneuromorphic design space explorationspiking neural networksedge computingquantization sensitivity
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

Energy-efficient edge intelligence needs simulators that capture how analog synapses and neurons actually behave, not only ideal spike models. This paper presents an open-source hardware-aware framework that places experimentally calibrated floating-gate and ReRAM synapse equations, plus several mixed-signal neuron circuits, inside custom PyTorch modules. Training therefore updates physical parameters such as floating-gate voltage or filament gap rather than abstract digital weights. On N-MNIST, DVS Gesture and Spiking Heidelberg Digits the tool returns classification accuracy together with estimated silicon area, power and quantization sensitivity, letting designers compare neuron, synapse and architecture choices against concrete hardware constraints before fabrication.

Core claim

By embedding nonlinear, measurement-calibrated floating-gate and ReRAM synapse models and multiple mixed-signal neuron implementations directly into a PyTorch training and inference loop, the framework enables end-to-end optimization of physical synaptic parameters and quantitative cross-layer design-space exploration of accuracy, area, power and quantization effects on standard neuromorphic benchmarks.

What carries the argument

Custom PyTorch modules that implement the modified EKV floating-gate current law and the exponential-sinh ReRAM I-V characteristic so the trainable weights are the device parameters V_FG0 and filament gap g; surrogate-gradient BPTT then updates those physical parameters while Euler-discretized neuron models (Axon-Hillock, LIF variants, Hodgkin-Huxley) produce spikes.

Load-bearing premise

That average-behavior device models and simple point-neuron circuits fitted to measured I-V and spike-rate curves are faithful enough to predict real silicon accuracy, area and power without modeling noise, mismatch, parasitics, routing delays or on-chip calibration.

What would settle it

Build a small mixed-signal SNN with the same FG or ReRAM synapses and neuron circuits, load the tool-optimized physical parameters with no extra tuning, and check whether measured classification accuracy, power and area on one benchmark match the framework's reported numbers within the claimed margins.

Watch this falsifier — get emailed when new claim-graph text bears on it.

If this is right

  • Designers can rank floating-gate versus ReRAM and LIF versus Axon-Hillock or Hodgkin-Huxley configurations by joint accuracy-area-power metrics before tape-out.
  • Post-training 8-bit floating-gate and 3-bit ReRAM quantization can be scored for accuracy drop without a separate abstract-to-device mapping step.
  • Recurrent and fully connected architectures can be compared under identical hardware models on temporal tasks such as SHD.
  • Neuron parameters extracted from both 65 nm and 28 nm processes can be swapped inside the same training pipeline, supporting multi-node exploration.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • If mean-device models prove sufficient, co-training on physical parameters could cut the lengthy bias and weight calibration that currently limits mixed-signal neuromorphic chips.
  • Adding device-noise and mismatch distributions to the same embedding would turn the explorer into a variability-aware optimizer rather than a mean-behavior tool.
  • An open common harness of this form could become a shared benchmark for comparing future analog synapse technologies on identical SNN tasks and metrics.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

4 major / 7 minor

Summary. The manuscript presents an open-source PyTorch framework for mixed-signal SNN design-space exploration that embeds experimentally calibrated floating-gate (65 nm) and ReRAM (Skywater 130 nm) nonlinear synapse models, together with multiple mixed-signal neuron models (Axon-Hillock, adaptive/standard/Schmitt LIF, Hodgkin–Huxley), into end-to-end training. Synaptic parameters optimized during learning are physical device quantities (V_FG0, filament gap g) rather than abstract weights. The tool supports fully connected and recurrent architectures, is evaluated on N-MNIST, DVS Gesture and SHD, and reports classification accuracy jointly with estimated silicon area, power and post-training quantization sensitivity. A small 2×8×2 XOR network is cross-checked against Cadence (function, area, power), and single-neuron ISI curves are matched to transistor-level simulations.

Significance. If the framework’s accuracy–area–power rankings are accepted as indicative of mixed-signal hardware trade-offs, this is a practically useful contribution: it lowers the barrier between algorithmic SNN work and analog neuromorphic circuit constraints, ships open code, and unifies several calibrated device and neuron models under one training loop. Strengths include measurement-anchored FG and ReRAM I–V models, Cadence-validated neuron ISI behavior, an explicit XOR circuit co-simulation, and joint reporting of accuracy with hardware metrics on standard neuromorphic datasets. The main value is as a co-design exploration platform rather than as a new learning algorithm or a fully predictive silicon model.

major comments (4)
  1. The central claim that the framework supports quantitative, hardware-predictive design-space exploration (Abstract; §1; §7) rests on mean-behavior models whose network-level fidelity is only weakly validated. Single-neuron ISI matches Cadence (Fig. 9–10; Supp. S3) and a 2×8×2 XOR net matches area/power to ~0.1% with 100% function (§6.1). Full-network results in Tables 2–4 are pure Python: no multi-neuron Cadence co-simulation, no injected mismatch/noise/parasitics/drift, and no AER/routing. §8 correctly flags these gaps, but the abstract and conclusion still present accuracy drops and ReRAM vs FG area/power rankings as hardware-oriented design guidance. Either temper the claim to “simulation-internal, mean-model exploration” or add at least one multi-neuron fidelity check (e.g., mismatch Monte Carlo or a larger Cadence/Python co-sim) that shows rankings are stable.
  2. Table 3 and §6: area and power are obtained by analytically scaling Cadence core-block layouts and accumulating branch/event currents × V_DD (Eq. 16). This omits interconnect, AER arbitration, memory access, and array parasitics that often dominate mixed-signal neuromorphic chips. The reported >100× area advantage of ReRAM vs FG and the absolute mW-scale numbers are therefore not yet comparable to deployable systems. Please state explicitly what is included/excluded in the area/power model, and either (i) add a sensitivity analysis under plausible routing/peripheral overheads or (ii) reframe Table 3 as core-compute estimates only, not full-chip metrics.
  3. §6.2 / Discussion: accuracy degradations of ~8–11% (N-MNIST) and ~7–8% (DVS) when replacing nn.Linear with ReRAM/FG are attributed to “hardware-induced non-idealities,” yet the authors note that disentangling nonlinearity, limited precision, and architecture is left for future work. Without an ablation (ideal linear weights vs nonlinear continuous device model vs post-training quantization alone; Tables 2 vs 4), the design-space conclusions about which neuron–synapse pair “best satisfies” accuracy–energy–area constraints are under-supported. A minimal ablation on one dataset would substantially strengthen the comparative claims.
  4. §3.1 and Table 4: quantization is applied post-training (8-bit FG, 3-bit ReRAM) by nearest-level projection after continuous optimization of V_FG0 or g. The framework’s stated advantage is optimizing physical parameters inside the training loop; discrete programmable levels are not part of that loop. For ReRAM especially (only 8 levels), PTQ can dominate the accuracy gap. Either integrate differentiable or STE-based quantization during training, or clearly separate “device-nonlinear continuous training” from “hardware-level discrete programming” when interpreting Table 4.
minor comments (7)
  1. Table 1 comparison is thin on modern hardware-aware or mixed-signal SNN tools (e.g., Brian2/Lava with device plugins, NeuroSim-class CIM estimators, SANA-FE cited only in limitations). A short positioning paragraph would help readers place the contribution.
  2. Fig. 1 caption and body: “eThe implementation” appears to be a typo; also “SnnTorch” / “snnTorch” capitalization is inconsistent.
  3. Eq. (2) text refers to Q_FG but the displayed equation uses capacitive coupling terms only; align notation with the prose.
  4. Table 2: architecture strings and neuron process nodes (65 nm vs 28 nm) are useful; please also report number of trainable device parameters and whether recurrent weights use the same FG/ReRAM model as feedforward weights.
  5. §5.1.2: Mutual SNN Pooling is important for DVS results; a one-sentence statement of whether pooling is fixed or learned, and whether it is counted in area/power, would avoid ambiguity.
  6. Code link (§9) is welcome; please pin a commit/tag and list which tables/figures are fully reproducible from the public repo.
  7. Supplementary temporal-deviation analysis (Fig. S3) is valuable; consider promoting a short quantitative summary (median |ΔV|, ISI error) into the main text near Fig. 9–10.

Circularity Check

0 steps flagged

No load-bearing circularity: device models are fitted to independent I–V/ISI measurements and Cadence baselines; network accuracy and area/power rankings are evaluated on external public datasets rather than reducing to those fits by construction.

full rationale

The paper’s central claim is an engineering framework that embeds calibrated FG (eqs. 1–2, Fig. 4) and ReRAM (eq. 3, Fig. 6) nonlinearities plus mixed-signal neuron models into PyTorch modules so that physical parameters (V_FG0, filament gap g) can be optimized end-to-end and accuracy/area/power/quantization reported jointly on N-MNIST, DVS Gesture and SHD. The device equations are ordinary empirical fits to measured I–V curves and Cadence ISI data; those fits are then used as fixed forward models inside training. Classification losses are computed on public external benchmarks, not on the same I–V or ISI data used for calibration, so the reported accuracies and the 8–11 % hardware-induced drops are not forced by construction. Area and power numbers are obtained by scaling separate Cadence layout baselines (Sec. 6.1 XOR match, Table 3). Self-citations ([37], [43], [59]) supply the measurement anchors and Verilog-A models; they are ordinary prior characterization work, not uniqueness theorems or load-bearing premises that close a definitional loop. No step equates a claimed prediction to its own fitted input, renames a known result, or smuggles an ansatz that forces the design-space rankings. The acknowledged fidelity gaps (noise, mismatch, AER, §8) are correctness risks, not circularity. Hence only a minimal residual score for ordinary self-citation of the authors’ own device characterizations.

Axiom & Free-Parameter Ledger

5 free parameters · 5 axioms · 0 invented entities

The central claim rests on standard SNN training machinery plus device and circuit models whose free parameters are fitted to measurements/Cadence, not invented particles. No new physical entities are postulated; the contribution is the integrated tool and the empirical design-space numbers it produces under those fitted models.

free parameters (5)
  • FG EKV/device fit parameters (κ, σ, I_thpmos, C ratios, V_TP, etc.)
    Extracted by fitting Eq. 1–2 to 65 nm FG measured I–V curves; they define the nonlinear map from trainable V_FG0 to synaptic current.
  • ReRAM model constants I0, g0, V0 and filament bounds
    Calibration parameters in Eq. 3 from Skywater 130 nm measurements; set the nonlinear I(V,g) used during training and 3-bit quantization levels.
  • Neuron circuit parameters (C_mem, thresholds, leak/adapt/ref currents, time constants)
    Matched to Cadence 65 nm / 28 nm simulations so Python ISI matches transistor-level behavior; chosen per neuron type rather than derived from first principles.
  • Quantization bit-widths and clip ranges (FG 8-bit, ReRAM 3-bit)
    Chosen from prior device characterization claims of achievable precision; post-training nearest-level projection depends on these discrete grids.
  • Network hyperparameters (layer sizes, epochs, T time bins, learning setup)
    Hand-chosen per dataset/architecture in Table 2; affect reported accuracy and power/area via utilization.
axioms (5)
  • domain assumption Surrogate-gradient BPTT through non-differentiable spikes yields useful credit assignment for these nonlinear analog synapses.
    §5 adopts standard surrogate BPTT without proving optimality for FG/ReRAM nonlinearities; authors note specialized algorithms are out of scope.
  • domain assumption Mean device I–V behavior (ignoring device-to-device and cycle-to-cycle variation during training) is sufficient for design-space ranking.
    §3.1.2 explicitly models mean ReRAM behavior; mismatch/noise deferred to future work (§8).
  • domain assumption First-order Euler discretization of membrane ODEs at chosen Δt preserves spike rates well enough for classification metrics.
    Used for AH/LIF/HH updates (§3.2); supplementary shows residual V_mem edge deviations vs Cadence.
  • ad hoc to paper Area and average power can be estimated by scaling Cadence block layouts and accumulating branch/event currents × V_DD.
    §6 formulas (16) and analytical component scaling; not full P&R or measured chip power for the large nets.
  • standard math Standard math of chain rule and matrix multiplies for custom nn.Module forward maps.
    PyTorch autodiff through FG/ReRAM current functions (§3.1 code figures).

pith-pipeline@v1.1.0-grok45 · 26253 in / 3542 out tokens · 47088 ms · 2026-07-14T16:00:56.092491+00:00 · methodology

0 comments
read the original abstract

Energy-efficient neuromorphic computing at the edge requires simulation tools that can capture the non-ideal behavior of mixed-signal spiking neural network (SNN) hardware while supporting system-level design exploration. This work presents an open-source hardware-aware simulation framework for mixed-signal SNNs that enables comparative analysis across neuron, synapse and architecture choices. The framework supports multiple neuron models, including Leaky Integrate-and-Fire (LIF), Hodgkin-Huxley (HH) and Axon-Hillock (AH), together with non-volatile analog synapses based on floating-gate transistors and ReRAM devices. By incorporating device-level nonlinearities directly into PyTorch-based training and inference, the tool enables optimization of physical synaptic parameters rather than idealized abstract weights. The framework is evaluated on standard neuromorphic benchmarks, including N-MNIST, DVS Gesture and Spiking Heidelberg Digits (SHD). For each model dataset configuration, it reports classification accuracy together with hardware-oriented metrics such as silicon area, power consumption and quantization sensitivity. These capabilities enable cross-layer design space exploration and help identify neuron-synapse configurations that best satisfy application-specific constraints on accuracy, energy efficiency, area and hardware fidelity.

Figures

Figures reproduced from arXiv: 2607.06456 by Aishwarya Natarajan, Corey Hart, Sahil Shah, Sayma Nowshin Chowdhury, Taseen Forhad, Vineeta Nair.

Figure 1
Figure 1. Figure 1: Impact of hardware non-idealities on neuromorphic learning and inference: [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: General flow of the proposed simulation framework for mixed-signal SNNs. [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Dynamics of representative spiking neuron models: Input spikes are [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: The figure shows the current vs input voltage relationships of an FG [PITH_FULL_IMAGE:figures/full_fig_p008_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Figure illustrating the python code for defining a custom pytorch class and [PITH_FULL_IMAGE:figures/full_fig_p009_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Current-voltage (I-V) characteristics of the ReRAM device. The ReRAM [PITH_FULL_IMAGE:figures/full_fig_p011_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: A demonstration of Python code that defines a custom PyTorch class and [PITH_FULL_IMAGE:figures/full_fig_p012_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: (a) Axon hillock circuit (b) LIF neuron circuit [PITH_FULL_IMAGE:figures/full_fig_p013_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Comparison of ISI vs. synaptic current between the Python model and [PITH_FULL_IMAGE:figures/full_fig_p015_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: (a) Hodgkin-Huxley neuron model (b) ISI vs. synaptic current for HH [PITH_FULL_IMAGE:figures/full_fig_p016_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: (a) Fully connected architecture (b) Recurrent connected architecture [PITH_FULL_IMAGE:figures/full_fig_p017_11.png] view at source ↗
Figure 12
Figure 12. Figure 12: Visualization of DVS Gesture dataset events spatial downsampling from [PITH_FULL_IMAGE:figures/full_fig_p020_12.png] view at source ↗
Figure 13
Figure 13. Figure 13: XOR Logic Decoding using Spiking Neural Activity in 65 nm [PITH_FULL_IMAGE:figures/full_fig_p021_13.png] view at source ↗
Figure 14
Figure 14. Figure 14: FG synapse-based SNN architecture for SHD dataset classification. (a) [PITH_FULL_IMAGE:figures/full_fig_p023_14.png] view at source ↗

Review history (2 revisions) →

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

61 extracted references · 5 linked inside Pith

  1. [1]

    Edge intelligence through in-sensor and near-sensor computing for the artificial intelligence of things,

    Y. Baek, B. Bae, H. Shin, C. Sonnadara, H. Cho, C.-Y. Lin, Y. Mu, C. Shen, S. Shah, G. Wang, and K. Lee, “Edge intelligence through in-sensor and near-sensor computing for the artificial intelligence of things,” npj Unconventional Computing , vol. 2, no. 1, p. 25, Oct. 2025. [Online]. A vailable:https://www.nature.com/articles/s44335-025-00040-6

    2025

  2. [2]

    Networks of spiking neurons: the third generation of neural network models,

    W. Maass, “Networks of spiking neurons: the third generation of neural network models,” Neural Networks, vol. 10, no. 9, pp. 1659–1671, 1997

    1997

  3. [3]

    A Survey of Neuromorphic Computing and Neural Networks in Hardware,

    C. D. Schuman, T. E. Potok, R. M. Patton, J. D. Birdwell, M. E. Dean, G. S. Rose, and J. S. Plank, “A Survey of Neuromorphic Computing and Neural Networks in Hardware,” May 2017, arXiv:1705.06963 [cs.NE]. [Online]. A vailable: http://arxiv.org/abs/1705.06963

    Pith/arXiv arXiv 2017

  4. [4]

    Loihi: A Neuromorphic Manycore Processor with On-Chip Learning,

    M. Davies, N. Srinivasa, T.-H. Lin, G. Chinya, Y. Cao, S. H. Choday, G. Dimou, P. Joshi, N. Imam, S. Jain, Y. Liao, C.-K. Lin, A. Lines, R. Liu, D. Mathaikutty, S. McCoy, A. Paul, J. Tse, G. Venkataramanan, Y.-H. Weng, A. Wild, Y. Yang, and H. Wang, “Loihi: A Neuromorphic Manycore Processor with On-Chip Learning,” IEEE Micro, vol. 38, no. 1, pp. 82–99, Ja...

    2018

  5. [5]

    TrueNorth: Design and Tool Flow of a 65 mW 1 Million Neuron Programmable Neurosynaptic Chip,

    F. Akopyan, J. Sawada, A. Cassidy, R. Alvarez-Icaza, J. Arthur, P. Merolla, N. Imam, Y. Nakamura, P. Datta, G.-J. Nam, B. Taba, M. Beakes, B. Brezzo, J. B. Kuang, R. Manohar, W. P. Risk, B. Jackson, and D. S. Modha, “TrueNorth: Design and Tool Flow of a 65 mW 1 Million Neuron Programmable Neurosynaptic Chip,” IEEE Transactions on Computer-Aided Design of ...

    2015

  6. [6]

    SpiNNaker: A 1-W 18-Core System-on-Chip for Massively-Parallel Neural Network Simulation,

    E. Painkras, L. A. Plana, J. Garside, S. Temple, F. Galluppi, C. Patterson, D. R. Lester, A. D. Brown, and S. B. Furber, “SpiNNaker: A 1-W 18-Core System-on-Chip for Massively-Parallel Neural Network Simulation,” IEEE Journal of Solid-State Circuits , vol. 48, no. 8, pp. 1943–1953, Aug. 2013, conference Name: IEEE Journal of Solid-State Circuits

    1943

  7. [7]

    A 65k-neuron 73-Mevents/s 22-pJ/event asynchronous micro-pipelined integrate-and-fire array transceiver,

    J. Park, S. Ha, T. Yu, E. Neftci, and G. Cauwenberghs, “A 65k-neuron 73-Mevents/s 22-pJ/event asynchronous micro-pipelined integrate-and-fire array transceiver,” in 2014 IEEE Biomedical Circuits and Systems Conference (BioCAS) Proceedings , Oct. 2014, pp. 675–678

    2014

  8. [8]

    Spike timing–dependent plasticity: a hebbian learning rule,

    N. Caporale and Y. Dan, “Spike timing–dependent plasticity: a hebbian learning rule,” Annu. Rev. Neurosci., vol. 31, no. 1, pp. 25–46, 2008

    2008

  9. [9]

    Neuromorphic electronic systems,

    C. Mead, “Neuromorphic electronic systems,” Proceedings of the IEEE , vol. 78, no. 10, pp. 1629– 1636, Oct. 1990, conference Name: Proceedings of the IEEE

    1990

  10. [10]

    Memory and Information Processing in Neuromorphic Systems,

    G. Indiveri and S.-C. Liu, “Memory and Information Processing in Neuromorphic Systems,” Proceedings of the IEEE , vol. 103, no. 8, pp. 1379–1397, Aug. 2015, conference Name: Proceedings of the IEEE

    2015

  11. [11]

    A reconfigurable on-line learning spiking neuromorphic processor comprising 256 neurons and 128k synapses,

    N. Qiao, H. Mostafa, F. Corradi, M. Osswald, F. Stefanini, D. Sumislawska, and G. Indiveri, “A reconfigurable on-line learning spiking neuromorphic processor comprising 256 neurons and 128k synapses,” Frontiers in neuroscience, vol. 9, p. 141, 2015. Hardware-Aware SNN Design Framework 28

    2015

  12. [12]

    Exploiting inherent error resiliency of deep neural networks to achieve extreme energy efficiency through mixed-signal neurons,

    B. Chatterjee, P. Panda, S. Maity, A. Biswas, K. Roy, and S. Sen, “Exploiting inherent error resiliency of deep neural networks to achieve extreme energy efficiency through mixed-signal neurons,” IEEE Transactions on Very Large Scale Integration (VLSI) Systems , vol. 27, no. 6, pp. 1365–1377, 2019

    2019

  13. [13]

    Neuromorphic analog circuits for robust on-chip always-on learning in spiking neural networks,

    A. Rubino, M. Cartiglia, M. Payvand, and G. Indiveri, “Neuromorphic analog circuits for robust on-chip always-on learning in spiking neural networks,” arXiv preprint arXiv:2307.06084 , 2023

    Pith/arXiv arXiv 2023

  14. [14]

    Variation-tolerant hardware synthesis of spiking neural networks,

    V. V Nair, S. N. Chowdhury, and S. Shah, “Variation-tolerant hardware synthesis of spiking neural networks,” in 2025 First International Conference on Intelligent Computing and Systems at the Edge (ICEdge) , vol. 1, 2025, pp. 1–6

    2025

  15. [15]

    Training Spiking Neural Networks Using Lessons From Deep Learning,

    J. K. Eshraghian, M. Ward, E. Neftci, X. Wang, G. Lenz, G. Dwivedi, M. Bennamoun, D. S. Jeong, and W. D. Lu, “Training Spiking Neural Networks Using Lessons From Deep Learning,” Aug. 2023, arXiv:2109.12894 [cs]. [Online]. A vailable: http://arxiv.org/abs/2109.12894

    Pith/arXiv arXiv 2023

  16. [16]

    Shaping the learning landscape in neural networks around wide flat minima,

    C. Baldassi, F. Pittorino, and R. Zecchina, “Shaping the learning landscape in neural networks around wide flat minima,” Proceedings of the National Academy of Sciences , vol. 117, no. 1, pp. 161–170, 2020

    2020

  17. [17]

    Gerstner, W

    W. Gerstner, W. M. Kistler, R. Naud, and L. Paninski, Neuronal Dynamics: From Single Neurons to Networks and Models of Cognition . Cambridge University Press, 2014

    2014

  18. [18]

    Genesis: A system for simulating neural networks,

    M. Wilson, U. Bhalla, J. Uhley, and J. Bower, “Genesis: A system for simulating neural networks,” Advances in neural information processing systems , vol. 1, 1988

    1988

  19. [19]

    The neuron simulation environment,

    M. L. Hines and N. T. Carnevale, “The neuron simulation environment,” Neural Computation, vol. 9, no. 6, pp. 1179–1209, 08 1997. [Online]. A vailable: https://doi.org/10.1162/neco.1997.9.6.1179

    doi:10.1162/neco.1997.9.6.1179 1997

  20. [20]

    Ermentrout, Simulating, Analyzing, and Animating Dynamical Systems: A Guide to XPPAUT for Researchers and Students

    B. Ermentrout, Simulating, Analyzing, and Animating Dynamical Systems: A Guide to XPPAUT for Researchers and Students . SIAM Press, 2002

    2002

  21. [21]

    Csim: A neural simulation environment for modeling the dynamics of spiking neuron networks,

    T. Natschläger, T. Bertschinger, R. Legenstein, and N. Maass, “Csim: A neural simulation environment for modeling the dynamics of spiking neuron networks,” Technical Report, 2003

    2003

  22. [22]

    Exact simulation of integrate-and-fire models with synaptic conductances,

    R. Brette, “Exact simulation of integrate-and-fire models with synaptic conductances,” Neural Computation, vol. 18, no. 8, pp. 2004–2027, 2006

    2004

  23. [23]

    Nest (neural simulation tool),

    M.-O. Gewaltig and M. Diesmann, “Nest (neural simulation tool),” Scholarpedia, vol. 2, no. 4, p. 1430, 2007

    2007

  24. [24]

    On the performance of voltage stepping for the simulation of adaptive, nonlinear integrate-and-fire neuronal networks,

    M. G. Kaabi, A. Tonnelier, and D. Martinez, “On the performance of voltage stepping for the simulation of adaptive, nonlinear integrate-and-fire neuronal networks,” Neural computation , vol. 23, no. 5, pp. 1187–1204, 2011

    2011

  25. [25]

    A large-scale model of the functioning brain,

    C. Eliasmith, T. C. Stewart, X. Choo, T. Bekolay, T. DeWolf, Y. Tang, and D. Rasmussen, “A large-scale model of the functioning brain,” Science, vol. 338, no. 6111, pp. 1202–1205, 2012

    2012

  26. [26]

    Ncs: A framework for simulating large- scale, biologically realistic neural networks,

    R. V. Hoang, F. C. H. Jr., P. H. Goodman, and M. Zirpe, “Ncs: A framework for simulating large- scale, biologically realistic neural networks,” in Proceedings of the International Joint Conference on Neural Networks (IJCNN) , 2003, pp. 411–416

    2003

  27. [27]

    Mosaic: in-memory computing and routing for small-world spike-based neuromorphic systems,

    T. Dalgaty, F. Moro, Y. Demirağ, A. De Pra, G. Indiveri, E. Vianello, and M. Payvand, “Mosaic: in-memory computing and routing for small-world spike-based neuromorphic systems,” Nature Communications, vol. 15, no. 1, p. 142, 2024

    2024

  28. [28]

    The neuron book,

    N. T. Carnevale and M. L. Hines, “The neuron book,” Cambridge University Press , 2006

    2006

  29. [29]

    Simulation of networks of spiking neurons: A review of tools and strategies,

    R. Brette, M. Rudolph, T. Carnevale, M. Hines, D. Beeman, J. M. Bower, M. Diesmann, A. Morrison, P. H. Goodman, F. C. H. Jr., M. Zirpe, T. Natschläger, D. Pecevski, B. Ermentrout, A. Djurfeldt, A. Lansner, O. Rochel, T. Vieville, E. Muller, A. P. Davison, S. H. B. Rotter, and A. Destexhe, “Simulation of networks of spiking neurons: A review of tools and s...

    2007

  30. [30]

    A floating gate and its application to memory devices,

    D. Kahng and S. M. Sze, “A floating gate and its application to memory devices,” The Bell System Technical Journal, vol. 46, no. 6, pp. 1288–1295, 1967

    1967

  31. [31]

    A floating-gate MOS learning array with locally computed weight updates,

    C. Diorio, P. Hasler, B. A. Minch, and C. A. Mead, “A floating-gate MOS learning array with locally computed weight updates,” IEEE Transactions on Electron Devices , vol. 44, no. 12, pp. Hardware-Aware SNN Design Framework 29 2281–2289, Dec. 1997, conference Name: IEEE Transactions on Electron Devices

    1997

  32. [32]

    Floating Gate Synapses With Spike-Time- Dependent Plasticity,

    S. Ramakrishnan, P. E. Hasler, and C. Gordon, “Floating Gate Synapses With Spike-Time- Dependent Plasticity,” IEEE Transactions on Biomedical Circuits and Systems , vol. 5, no. 3, pp. 244–252, Jun. 2011, conference Name: IEEE Transactions on Biomedical Circuits and Systems

    2011

  33. [33]

    Temporally learning floating-gate VLSI synapses,

    S.-C. Liu and R. Mockel, “Temporally learning floating-gate VLSI synapses,” in 2008 IEEE International Symposium on Circuits and Systems , May 2008, pp. 2154–2157

    2008

  34. [34]

    Implementation of multilayer perceptron network with highly uniform passive memristive crossbar circuits,

    F. M. Bayat, M. Prezioso, B. Chakrabarti, H. Nili, I. Kataeva, and D. Strukov, “Implementation of multilayer perceptron network with highly uniform passive memristive crossbar circuits,” Nature Communications , vol. 9, no. 1, p. 2331, Jun. 2018, number: 1. [Online]. A vailable: https://www.nature.com/articles/s41467-018-04482-4

    2018

  35. [35]

    An analytical MOS transistor model valid in all regions of operation and dedicated to low-voltage and low-current applications,

    C. C. Enz, F. Krummenacher, and E. A. Vittoz, “An analytical MOS transistor model valid in all regions of operation and dedicated to low-voltage and low-current applications,” Analog Integrated Circuits and Signal Processing , vol. 8, no. 1, pp. 83–114, Jul. 1995. [Online]. A vailable:https://doi.org/10.1007/BF01239381

    doi:10.1007/bf01239381 1995

  36. [36]

    Modeling, simulation and implementation of circuit elements in an open-source tool set on the FPAA,

    A. Natarajan and J. Hasler, “Modeling, simulation and implementation of circuit elements in an open-source tool set on the FPAA,” Analog Integrated Circuits and Signal Processing , vol. 91, no. 1, pp. 119–130, Apr. 2017. [Online]. A vailable: http://link.springer.com/10.1007/s10470-016-0914-y

    doi:10.1007/s10470-016-0914-y 2017

  37. [37]

    Analysis and verilog-a modeling of floating-gate transistors,

    S. N. Chowdhury, M. Chen, and S. Shah, “Analysis and verilog-a modeling of floating-gate transistors,” IEEE Open Journal of Circuits and Systems , 2024

    2024

  38. [38]

    Integrated Floating-Gate Programming Environment for System-Level ICs,

    S. Kim, J. Hasler, and S. George, “Integrated Floating-Gate Programming Environment for System-Level ICs,” IEEE Transactions on Very Large Scale Integration (VLSI) Systems , vol. 24, no. 6, pp. 2244–2252, Jun. 2016, conference Name: IEEE Transactions on Very Large Scale Integration (VLSI) Systems

    2016

  39. [39]

    Modeling and analysis of analog non-volatile devices for compute-in-memory applications,

    C. Brando, M. Park, S. N. Chowdhury, M. Chen, K. Lee, and S. Shah, “Modeling and analysis of analog non-volatile devices for compute-in-memory applications,” in 2023 IEEE 66th International Midwest Symposium on Circuits and Systems (MWSCAS) . IEEE, 2023, pp. 103–107

    2023

  40. [40]

    10 × 10nm 2 hf/hfo x crossbar resistive ram with excellent performance, reliability and low-energy operation,

    B. Govoreanu, G. S. Kar, Y. Chen, V. Paraschiv, S. Kubicek, A. Fantini, I. Radu, L. Goux, S. Clima, R. Degraeve et al. , “10 × 10nm 2 hf/hfo x crossbar resistive ram with excellent performance, reliability and low-energy operation,” in 2011 International Electron Devices Meeting. IEEE, 2011, pp. 31–6

    2011

  41. [41]

    Influence of oxygen vacancies in ald hfo2-x thin films on non-volatile resistive switching phenomena with a ti/hfo2-x/pt structure,

    A. S. Sokolov, Y.-R. Jeon, S. Kim, B. Ku, D. Lim, H. Han, M. G. Chae, J. Lee, B. G. Ha, and C. Choi, “Influence of oxygen vacancies in ald hfo2-x thin films on non-volatile resistive switching phenomena with a ti/hfo2-x/pt structure,” Applied Surface Science, vol. 434, pp. 822–830, 2018

    2018

  42. [42]

    A low-temperature-grown tio2-based device for the flexible stacked rram application,

    H. Y. Jeong, Y. I. Kim, J. Y. Lee, and S.-Y. Choi, “A low-temperature-grown tio2-based device for the flexible stacked rram application,” Nanotechnology, vol. 21, no. 11, p. 115203, 2010

    2010

  43. [43]

    Characterization and modeling of multilevel analog reram synapses in the sky130 process,

    I. Didin, C. Brando, C.-Y. Lin, and S. Shah, “Characterization and modeling of multilevel analog reram synapses in the sky130 process,” IEEE Journal on Exploratory Solid-State Computational Devices and Circuits , 2026

    2026

  44. [44]

    Neuromorphic Silicon Neuron Circuits,

    G. Indiveri, B. Linares-Barranco, T. J. Hamilton, A. v. Schaik, R. Etienne-Cummings, T. Delbruck, S.-C. Liu, P. Dudek, P. Häfliger, S. Renaud, J. Schemmel, G. Cauwenberghs, J. Arthur, K. Hynna, F. Folowosele, S. Saighi, T. Serrano-Gotarredona, J. Wijekoon, Y. Wang, and K. Boahen, “Neuromorphic Silicon Neuron Circuits,” Frontiers in Neuroscience, vol. 5, 2011

    2011

  45. [45]

    Mead, Analog VLSI and neural systems

    C. Mead, Analog VLSI and neural systems . USA: Addison-Wesley Longman Publishing Co., Inc., 1989

    1989

  46. [46]

    A low-power adaptive integrate-and-fire neuron circuit,

    G. Indiveri, “A low-power adaptive integrate-and-fire neuron circuit,” in Proceedings of the 2003 International Symposium on Circuits and Systems, 2003. ISCAS ’03. , vol. 4, 2003, pp. IV–IV

    2003

  47. [47]

    Hardware aware modeling of mixed-signal spiking neural network,

    S. Chowdhury and S. Shah, “Hardware aware modeling of mixed-signal spiking neural network,” IEEE NEWCAS , 2022

    2022

  48. [48]

    A quantitative description of membrane current and its application to conduction and excitation in nerve,

    A. L. Hodgkin and A. F. Huxley, “A quantitative description of membrane current and its application to conduction and excitation in nerve,” The Journal of Physiology , vol. 117, no. 4, pp. 500–544, 1952. [Online]. A vailable: https://physoc.onlinelibrary.wiley.com/doi/abs/10.1113/jphysiol.1952.sp004764

    doi:10.1113/jphysiol.1952.sp004764 1952

  49. [49]

    Hodgkin–Huxley Neuron and FPAA Dynamics,

    A. Natarajan and J. Hasler, “Hodgkin–Huxley Neuron and FPAA Dynamics,” IEEE Transactions on Biomedical Circuits and Systems , vol. 12, no. 4, pp. 918–926, Aug. 2018, conference Name: IEEE Transactions on Biomedical Circuits and Systems

    2018

  50. [50]

    Backpropagation through time and the brain,

    T. P. Lillicrap and A. Santoro, “Backpropagation through time and the brain,” Current opinion in neurobiology, vol. 55, pp. 82–89, 2019

    2019

  51. [51]

    Surrogate Gradient Learning in Spiking Neural Networks,

    E. O. Neftci, H. Mostafa, and F. Zenke, “Surrogate Gradient Learning in Spiking Neural Networks,” arXiv:1901.09948 [cs, q-bio] , May 2019, arXiv: 1901.09948. [Online]. A vailable: http://arxiv.org/abs/1901.09948

    Pith/arXiv arXiv 1901

  52. [52]

    Converting Static Image Datasets to Spiking Neuromorphic Datasets Using Saccades,

    G. Orchard, A. Jayawant, G. K. Cohen, and N. Thakor, “Converting Static Image Datasets to Spiking Neuromorphic Datasets Using Saccades,” Frontiers in Neuroscience , vol. 9, 2015. [Online]. A vailable: https://www.frontiersin.org/articles/10.3389/fnins.2015.00437/full

    doi:10.3389/fnins.2015.00437/full 2015

  53. [53]

    A low power, fully event-based gesture recognition system,

    A. Amir, B. Taba, D. Berg, T. Melano, J. McKinstry, C. Di Nolfo, T. Nayak, A. Andreopoulos, G. Garreau, M. Mendoza et al. , “A low power, fully event-based gesture recognition system,” in Proceedings of the IEEE conference on computer vision and pattern recognition , 2017, pp. 7243–7252

    2017

  54. [54]

    The heidelberg spiking data sets for the systematic evaluation of spiking neural networks,

    B. Cramer, Y. Stradmann, J. Schemmel, and F. Zenke, “The heidelberg spiking data sets for the systematic evaluation of spiking neural networks,” IEEE Transactions on Neural Networks and Learning Systems, vol. 33, no. 7, pp. 2744–2757, 2022

    2022

  55. [55]

    Deep learning for event-based vision: A comprehensive survey and benchmarks,

    X. Zheng, Y. Liu, Y. Lu, T. Hua, T. Pan, W. Zhang, D. Tao, and L. Wang, “Deep learning for event-based vision: A comprehensive survey and benchmarks,” arXiv preprint arXiv:2302.08890, 2023

    Pith/arXiv arXiv 2023

  56. [56]

    Tonic: event-based datasets and transformations

    G. Lenz, K. Chaney, S. Bam Shrestha, O. Oubari, S. Picaud, and G. Zarrella, “Tonic: event-based datasets and transformations. ” Zenodo, 2021

    2021

  57. [57]

    Performance comparison of dvs data spatial downscaling methods using spiking neural networks,

    A. Gruel, J. Martinet, B. Linares-Barranco, and T. Serrano-Gotarredona, “Performance comparison of dvs data spatial downscaling methods using spiking neural networks,” in Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision , 2023, pp. 6494–6502

    2023

  58. [58]

    Sana-fe: Simulating advanced neuromorphic architectures for fast exploration,

    J. A. Boyle, M. Plagge, S. G. Cardwell, F. S. Chance, and A. Gerstlauer, “Sana-fe: Simulating advanced neuromorphic architectures for fast exploration,” IEEE Transactions on Computer- Aided Design of Integrated Circuits and Systems , vol. 44, no. 8, pp. 3165–3178, 2025

    2025

  59. [59]

    Open-source floating-gate cell for analogue synapses,

    M. Chen, C. Sonnadara, and S. Shah, “Open-source floating-gate cell for analogue synapses,” Electronics Letters , vol. 60, no. 17, p. e70036, 2024, _eprint: https://onlinelibrary.wiley.com/doi/pdf/10.1049/ell2.70036. [Online]. A vailable: https://onlinelibrary.wiley.com/doi/abs/10.1049/ell2.70036 Supplementary Material Supplementary Material 31

    doi:10.1049/ell2.70036 2024

  60. [60]

    Standard LIF neuron M84M1 M M3M2 C1 M5 M6 M9 M7 VspikeIleak Iin Figure S1: Standard LIF neuron circuit used in the manuscript The LIF neuron circuit, shown in Fig

    Neuron dynamics 11.1. Standard LIF neuron M84M1 M M3M2 C1 M5 M6 M9 M7 VspikeIleak Iin Figure S1: Standard LIF neuron circuit used in the manuscript The LIF neuron circuit, shown in Fig. S1, uses a membrane capacitor ( C1) to integrate input current ( Iin). Its temporal dynamics are governed by: C1 dVmem dt = −gl ( Vmem − Vreset ) + Iin(t) (S17) where gl r...

  61. [61]

    Quantization 12.1. Floating gate synapse This work implements a hardware-grounded 8-bit post-training quantization (PTQ) scheme rooted in the floating-gate synapse characterization frameworks established by [ 37, 59]. As characterized, the physical device exhibits threshold voltage ( Vtp) levels distributed across a 0–5 V dynamic range with device-level v...