Pith. sign in

REVIEW 4 major objections 5 minor 31 references

FFGAF-SNN: The Forward-Forward Based Gradient Approximation Free Training Framework for Spiking Neural Networks

T0 review · 4 major / 5 minor · reviewed 2026-08-06 · deepseek-v4-flash

Pith's one-line read A gradient-free, forward-only training rule brings convolutional spiking networks to 99.58% accuracy on MNIST and 75.64% on CIFAR-10.

desk verdict Plausible FF-SNN recipe with a load-bearing gap: the channel-allocation features are unspecified, so the headline margins may not survive a clean re-run. read the letter →

arxiv 2507.23643 v2 pith:NWBWX34U submitted 2025-07-31 cs.CV

classification cs.CV
keywords Forward-Forwardalgorithmspikingneuralnetworkssurrogate-gradient-freetrainingblack-boxlayerschannel-wisecompetitivelearningclass-awarecomplexityANN-SNNconversionenergy-efficient
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

The paper claims that spiking neural networks can be trained without backpropagation and without any surrogate-gradient approximation, by combining the Forward-Forward algorithm with convolutional layers and freezing the spiking neuron layers as non-trainable black-box encoders. Training then only updates the convolutional feature extractors with local goodness losses, while the spikes are treated as opaque but informative encodings. The paper adds a class-aware channel-allocation mechanism that gives more channels to classes that are harder to tell apart, based on cosine similarities among class-mean feature vectors. On MNIST, Fashion-MNIST, and CIFAR-10, this reaches 99.58%, 92.13%, and 75.64% test accuracy, above all previously reported Forward-Forward training schemes for SNNs and close to backpropagation-trained SNNs, while using fewer parameters and less memory-access energy.

What carries the argument

The load-bearing mechanism is the black-box spiking layer: integrate-and-fire neurons are frozen and never receive weight updates, so their non-differentiable activation function never needs a gradient. Around this sits a channel-wise competitive loss, taken from prior Forward-Forward work, which partitions the channel dimension by class and measures each class's 'goodness' as the mean squared spatiotemporal activation; a locally normalized softmax-style cross-entropy loss trains each layer block independently. Two supporting components are an improved ReLU activation (a clipped, level-shifted function borrowed from ANN-SNN conversion) that reduces encoding error at the spike layer, and a regularization layer that rescales outputs by a trainable threshold to keep them inside the spiking neuron's expressible range. The allocation rule works by computing class-mean feature vectors, cosine similarities between classes, standardized similarity sums, and a proportion-based channel split with a uniformity hyperparameter fixed at 2.

What would settle it

Re-run the CIFAR-10 experiment with the channel-allocation scores recomputed from class-mean features of only the training split, produced by a randomly initialized, untrained convolutional network, and compare against both the reported 75.64% and the uniform-allocation baseline; if accuracy falls to the uniform-allocation level (about 75.28% for the comparable model), the allocation advantage came from information about the test classes rather than from the complexity measure itself.

Watch

Extended reading notes

Core claim

The central discovery is that the non-differentiability of spiking neurons does not have to be worked around at all: if the spiking layer is inserted as a frozen black box that encodes rectified, normalized activations into spike trains, a Forward-Forward training signal defined on the squared, spatiotemporal mean of per-class channel subsets can drive the trainable convolutional layers to competitive accuracy. The paper further asserts that distributing channels according to normalized inter-class similarity scores, rather than uniformly, improves accuracy under fixed parameter budgets because harder classes receive more representational resources. The method is claimed to be the first Forward-Forward training of convolutional spiking networks, and to reduce parameter counts to about 2.3% of a baseline MLP while lowering memory-access energy.

Load-bearing premise

The channel-allocation advantage rests entirely on the class-mean feature vectors in Eq. (1): the paper does not say which feature extractor produced them or whether the statistics were taken from the training split, the validation split, or the full dataset including test images, so if those vectors encode test knowledge, the reported accuracy gains over uniform allocation would not transfer to unseen data.

Editorial extensions

If this is right

  • Convolutional spiking networks can be trained end-to-end with purely forward passes, with no surrogate gradient, no backward pass, and no stored intermediate activations for backpropagation.
  • Assigning more channels to confusable classes improves accuracy when parameters are scarce; the gap over uniform allocation closes as the model's capacity grows, so the benefit appears exactly in the edge-deployment regime the paper targets.
  • The same frozen-black-box trick applies to any non-differentiable neuron model, so extending to richer spiking dynamics does not require new training mathematics.
  • At 433,600 parameters and 0.081 mJ total estimated energy per inference, the trained network is small enough for on-device learning if the Forward-Forward updates are implemented in hardware.
  • The reported accuracies beat every Forward-Forward-trained spiking network in the comparison table while using fewer epochs on MNIST and Fashion-MNIST than most baselines.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • If the class-mean features used for allocation (Eq. 1) are computed from the training split only with a feature extractor that does not already encode the target classes, the method becomes a fully data-driven architecture allocation rule; the paper leaves this provenance unstated, and the claimed allocation gains should be re-measured under that constraint.
  • Because allocation is computed once before training, an obvious extension the paper does not explore is to update the allocation during training as class-mean features drift, which might further raise the CIFAR-10 result where inter-class difficulty differences are small.
  • The black-box spiking layer suggests a modular co-design: any neuromorphic core that maps rectified inputs to spikes could be dropped into the architecture without altering the learning rule, connecting the method to heterogeneous hardware accelerators.
  • The energy comparison depends on theoretical estimates from a specific analytical model; the interesting test is whether these ratios hold on actual neuromorphic hardware, where memory-access energy and spike sparsity interact differently.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 5 minor

Summary. The manuscript proposes FFGAF-SNN, a Forward-Forward-based training framework for convolutional spiking neural networks. Spiking layers are frozen and treated as non-trainable black-box encoders, while trainable convolutional blocks are optimized by a cross-entropy-style goodness loss. A class-complexity analysis of dataset features (Eqs. 1-4) is used to allocate channels per class, and an improved ReLU/quantization activation (Eq. 13) plus output regularization is introduced to reduce spiking-layer conversion error. Experiments report test accuracies of 99.58% on MNIST, 92.13% on Fashion-MNIST, and 75.64% on CIFAR-10, claiming to surpass all existing FF-based SNN approaches and several non-backpropagation baselines, with lower parameter counts and estimated energy consumption.

Significance. If the reported results are reproducible, the work is a meaningful step: it appears to be the first demonstration of Forward-Forward-style training on convolutional spiking networks, and the frozen black-box treatment of spiking neurons is a clean way to avoid surrogate-gradient approximations. The comparison in Table 1 covers a broad set of FF, STDP, and direct-feedback-alignment baselines, and the parameter/energy analysis in Table 2 is useful. The central novelty, however, is the class-complexity channel allocation of Eq. (4), and its empirical support currently rests on an incompletely specified feature-extraction step. Because the claimed margins over the closest CNN baseline are small, the leakage concern is not a presentation detail but a threat to the headline comparison.

major comments (4)
  1. [Inter-class Similarity Analysis, Eqs. (1)-(4)] The feature extractor that produces the vectors f_i in Eq. (1), and the data split D used to compute the class means, are not specified. If D includes test images, or if the features come from a network trained on the target classes, then the channel allocation of Eq. (4) uses test information, and the reported accuracy gain over uniform allocation (e.g., 75.64% vs. 75.28% on CIFAR-10) would not transfer to unseen data. Please specify the extractor and the split, and rerun the comparison using statistics computed from the training split only.
  2. [Experiments and results, Table 1] All accuracies are single-run values with no standard deviations or error bars. Given that the headline comparison against CwComp* rests on differences below 0.5 percentage points on MNIST and CIFAR-10, it is impossible to judge whether the improvement is statistically meaningful. Report mean and standard deviation over at least five seeds for the proposed method and for the re-implemented baselines.
  3. [Loss Function, Eq. (6)] The loss is written as cross-entropy applied directly to goodness values G_n,j, but no softmax or normalization is defined, and G_n,j is a sum of squared activations rather than a probability. Because each class j is assigned disjoint channels, the positive-class term does not appear to compete with the other classes, so the objective may simply drive up the absolute goodness of the correct class without suppressing incorrect ones. Please clarify how this acts as a classification objective, or compare with a properly normalized softmax cross-entropy variant.
  4. [Tables 1 and 2] FF* and CwComp* are described as 'our implemented models', but no implementation details, hyperparameters, or code are provided, and the energy values in Table 2 are analytical estimates. This makes the comparison hard to verify and weakens the claim of surpassing all existing FF-based SNN approaches. Provide architecture details and training hyperparameters for the re-implemented baselines, and state the assumptions behind the energy model.
minor comments (5)
  1. [Inter-class Similarity Analysis, Eqs. (2) and (4)] The symbol S is used both for the class-similarity matrix S_{c,c'} and for the channel-allocation variable S_j, which is confusing; please use distinct notation.
  2. [Inter-class Similarity Analysis, Eq. (1)] The text says 'where D denotes the number of datasets', but D should denote the dataset itself; please correct this typo.
  3. [Table 1] The header 'Test Er. (%)' is inconsistent with the entries, which are accuracies rather than errors; also correct the typos 'Fashion-MIST' and 'Fashion-MINST'.
  4. [Ablation Study] The ablation section states that on CIFAR-10 the advantage of the proposed allocation is 'not fully manifested' because inter-class difficulty variations are slight; this tempers the abstract's claim that the mechanism is a key source of improvement and should be acknowledged in the conclusion.
  5. [Network, Eq. (13)] Equation (13) is called an improved ReLU function, but it is a clipped quantization function; the connection to ReLU and to conversion-error reduction should be stated more explicitly.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the reported accuracies are empirical outcomes of a forward-forward training pipeline, not quantities defined by the method's inputs.

full rationale

The claimed derivation chain is self-contained: spiking layers are frozen black-box encoders, the goodness score is computed by Eq. (5) from allocated channel subsets, the loss is the cross-entropy in Eq. (6), and the channel allocation in Eq. (4) is computed from dataset class-mean features via Eqs. (1)-(3). None of these equations is defined in terms of the final test accuracies, and no fitted parameter is renamed as a prediction. The hyperparameter phi=2 is fixed, not tuned to the reported results. The self-citations (Xu, Liu, and Yang 2023, 2024; Pei et al. 2023) appear only in the introductory motivation and are not load-bearing for the method or its comparison. External citations (Hinton 2022; Papachristodoulou et al. 2024; Bu et al. 2023; Zheng et al. 2021) supply the FF framing, channel partitioning, improved ReLU, and temporal normalization without being used to equate an output with an input. The main caveat, that Eq. (1) does not identify the feature extractor or data split used for the class-mean vectors, is a potential data-leakage/soundness concern rather than a circularity: the allocation is a precomputed design choice, not a post-hoc fit to the accuracies being reported. Therefore no circular step can be exhibited from the paper's equations, and the circularity score is 0.

Assumptions & free parameters 4 free parameters · 4 assumptions · 0 invented entities

The method introduces no new physical or mathematical entities. Its load-bearing assumptions are the FF local-learning premise, the IF neuron model for the black-box layer, the conversion-error-reducing property of the chosen ReLU variant, and the use of an unnormalized goodness value as a classification probability. The main free parameters are the channel-allocation hyperparameters, several of which are left underspecified.

free parameters (4)
  • phi (channel allocation uniformity) = 2
    Fixed hyperparameter in Eq. (4) governing the uniformity of channel allocation; set without a search procedure described.
  • thresh (regularized output scale)
    Scales the regularized output before the SNN layer (Eq. 10); no value or tuning procedure is reported.
  • L and lambda_l (quantization levels and scaling)
    Parameters in the improved ReLU (Eq. 13), inherited from Bu et al. 2023; values are not listed.
  • channel allocation S_j = data-derived
    Per-class channel counts computed from normalized class-complexity scores (Eq. 4); the feature extractor and data split behind these scores are unspecified.
assumptions (4)
  • domain assumption Forward-Forward local layer-wise objectives produce useful features for classification.
    The method relies on local goodness losses per block; this is the core premise of Hinton's FF algorithm, cited but not re-derived.
  • domain assumption The IF neuron model in Eq. (11) with reset describes the spiking black-box layer accurately enough.
    The conversion-error derivation assumes this membrane potential update; the paper does not justify it against other neuron models.
  • domain assumption The modified ReLU from Bu et al. reduces conversion error between analog activations and spike encodings.
    Eq. (13) is taken from Bu et al. 2023; the reduction in quantization error is cited, not re-derived here.
  • ad hoc to paper Cross-entropy applied to goodness values G_n,j (Eq. 6) is a valid training objective without an explicit softmax or normalization.
    As written, G_n,j is a mean-square statistic, not a probability; the loss is only a true cross-entropy if a softmax or normalization is applied somewhere, which the paper does not state.

how reviews work

0 comments
Cite this review

Pith. "Pith review of FFGAF-SNN: The Forward-Forward Based Gradient Approximation Free Training Framework for Spiking Neural Networks." pith.science (2026). https://pith.science/paper/NWBWX34U

@misc{pith2026250723643,
  author       = {Pith},
  title        = {Pith review of: FFGAF-SNN: The Forward-Forward Based Gradient Approximation Free Training Framework for Spiking Neural Networks},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/NWBWX34U}},
  note         = {Machine review of arXiv:2507.23643}
}
read the original abstract

Spiking Neural Networks (SNNs) offer a biologically plausible framework for energy-efficient neuromorphic computing. However, it is a challenge to train SNNs due to their non-differentiability, efficiently. Existing gradient approximation approaches frequently sacrifice accuracy and face deployment limitations on edge devices due to the substantial computational requirements of backpropagation. To address these challenges, we propose a Forward-Forward (FF) based gradient approximation-free training framework for Spiking Neural Networks, which treats spiking activations as black-box modules, thereby eliminating the need for gradient approximation while significantly reducing computational complexity. Furthermore, we introduce a class-aware complexity adaptation mechanism that dynamically optimizes the loss function based on inter-class difficulty metrics, enabling efficient allocation of network resources across different categories. Experimental results demonstrate that our proposed training framework achieves test accuracies of 99.58%, 92.13%, and 75.64% on the MNIST, Fashion-MNIST, and CIFAR-10 datasets, respectively, surpassing all existing FF-based SNN approaches. Additionally, our proposed method exhibits significant advantages in terms of memory access and computational power consumption.

Figures

Figures reproduced from arXiv: 2507.23643 by the authors.

Figure 2
Figure 2. The discriminative complexity of each class was [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 1
Figure 1. Class similarity heatmap of the CIFAR-10 dataset, [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 3
Figure 3. SNN network architecture based on the Forward-Forward (FF) algorithm. The network comprises an input layer, [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: (a)The figure demonstrates the accuracy variation [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]
Figure 5
Figure 5. Figure 5: CONV Feature Maps: Feature maps extracted by convolution, representing outputs across different channels.Weighted [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

31 extracted references · 18 canonical work pages

  1. [1]

    M.; Kok, J

    Bohte, S. M.; Kok, J. N.; and La Poutr \'e , J. A. 2000. SpikeProp: backpropagation for networks of spiking neurons. In ESANN, volume 48, 419--424. Bruges

  2. [2]

    Bu, T.; Fang, W.; Ding, J.; Dai, P.; Yu, Z.; and Huang, T. 2023. Optimal ANN-SNN conversion for high-accuracy and ultra-low-latency spiking neural networks. arXiv preprint arXiv:2303.04347

  3. [3]

    U.; and Cook, M

    Diehl, P. U.; and Cook, M. 2015. Unsupervised learning of digit recognition using spike-timing-dependent plasticity. Frontiers in computational neuroscience, 9: 99

  4. [4]

    U.; Neil, D.; Binas, J.; Cook, M.; Liu, S.-C.; and Pfeiffer, M

    Diehl, P. U.; Neil, D.; Binas, J.; Cook, M.; Liu, S.-C.; and Pfeiffer, M. 2015. Fast-classifying, high-accuracy spiking deep networks through weight and threshold balancing. In 2015 International joint conference on neural networks (IJCNN), 1--8. ieee

  5. [5]

    J.; and Oramas, J

    Dooms, T.; Tsang, I. J.; and Oramas, J. 2023. The trifecta: Three simple techniques for training deeper forward-forward networks. arXiv preprint arXiv:2311.18130

  6. [6]

    u gel, K.; Coquelin, D.; Weiel, M.; Debus, C.; Streit, A.; and G \

    Fl \"u gel, K.; Coquelin, D.; Weiel, M.; Debus, C.; Streit, A.; and G \"o tz, M. 2023. Feed-forward optimization with delayed feedback for neural networks. arXiv preprint arXiv:2304.13372

  7. [7]

    Frenkel, C.; Lefebvre, M.; and Bol, D. 2021. Learning without feedback: Fixed random learning signals allow for feedforward training of deep neural networks. Frontiers in neuroscience, 15: 629892

  8. [8]

    Backpropagation-free Spiking Neural Networks with the Forward-Forward Algorithm

    Ghader, M.; Kheradpisheh, S. R.; Farahani, B.; and Fazlali, M. 2025. Backpropagation-free Spiking Neural Networks with the Forward-Forward Algorithm. arXiv preprint arXiv:2502.20411

Show all 31 references
  1. [9]

    Hao, Y.; Huang, X.; Dong, M.; and Xu, B. 2020. A biologically plausible supervised learning method for spiking neural networks using the symmetric STDP rule. Neural Networks, 121: 387--395

  2. [10]

    Hinton, G. 2022. The forward-forward algorithm: Some preliminary investigations. arXiv preprint arXiv:2212.13345, 2(3): 5

  3. [11]

    Karkehabadi, A.; Homayoun, H.; and Sasan, A. 2024. FFCL: forward-forward net with cortical loops, training and inference on edge without Backpropogation. In proceedings of the Great Lakes symposium on VLSI 2024, 626--632

  4. [12]

    R.; Ganjtabesh, M.; Thorpe, S

    Kheradpisheh, S. R.; Ganjtabesh, M.; Thorpe, S. J.; and Masquelier, T. 2018. STDP-based spiking deep convolutional neural networks for object recognition. Neural Networks, 99: 56--67

  5. [13]

    Krizhevsky, A.; and Hinton, G. 2009. Learning multiple layers of features from tiny images. Technical report, University of Toronto

  6. [14]

    LeCun, Y.; Cortes, C.; and Burges, C. J. 1998. The MNIST database of handwritten digits

  7. [15]

    Lee, H.-C.; and Song, J. 2023. Symba: Symmetric backpropagation-free contrastive learning with forward-forward algorithm for optimizing convergence. arXiv preprint arXiv:2303.08418

  8. [16]

    Lemaire, E.; Cordone, L.; Castagnetti, A.; Novac, P.-E.; Courtois, J.; and Miramond, B. 2022. An analytical estimation of spiking neural networks energy efficiency. In International conference on neural information processing, 574--587. Springer

  9. [17]

    Lorberbom, G.; Gat, I.; Adi, Y.; Schwing, A.; and Hazan, T. 2024. Layer collaboration in the forward-forward algorithm. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 38, 14141--14148

  10. [18]

    Ororbia, A.; and Mali, A. 2023. The predictive forward-forward algorithm. arXiv preprint arXiv:2301.01452

  11. [19]

    Ororbia, A. G. 2024. Contrastive signal--dependent plasticity: Self-supervised learning in spiking neural circuits. Science Advances, 10(43): eadn6076

  12. [20]

    Papachristodoulou, A.; Kyrkou, C.; Timotheou, S.; and Theocharides, T. 2024. Convolutional channel-wise competitive learning for the forward-forward algorithm. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 38, 14536--14544

  13. [21]

    Pei, Y.; Xu, C.; Wu, Z.; Liu, Y.; and Yang, Y. 2023. ALBSNN: ultra-low latency adaptive local binary spiking neural network with accuracy loss estimator. Frontiers in Neuroscience, 17: 1225871

  14. [22]

    Taylor, L.; King, A.; and Harper, N. 2022. Robust and accelerated single-spike spiking neural network training with applicability to challenging temporal tasks. arXiv preprint arXiv:2205.15286

  15. [23]

    B.; Del Ser, J.; Mart \' nez-Seras, A.; and Garcia-Bringas, P

    Terres-Escudero, E. B.; Del Ser, J.; Mart \' nez-Seras, A.; and Garcia-Bringas, P. 2024. On the Robustness of Fully-Spiking Neural Networks in Open-World Scenarios using Forward-Only Learning Algorithms. arXiv e-prints, arXiv--2407

  16. [24]

    WANG, J.; WANG, J.-p.; SUN, W.-h.; and CHEN, S. 2020. A neuromorphic hardware design of a spiking convolutional neural network. Microelectronics & Computer, 37(12): 1--5

  17. [25]

    Xiao, H.; Rasul, K.; and Vollgraf, R. 2017. Fashion-MNIST: a novel image dataset for benchmarking machine learning algorithms. arXiv preprint arXiv:1708.07747

  18. [26]

    Xu, C.; Liu, Y.; and Yang, Y. 2023. Ultra-low latency spiking neural networks with spatio-temporal compression and synaptic convolutional block. Neurocomputing, 550: 126485

  19. [27]

    Xu, C.; Liu, Y.; and Yang, Y. 2024. STCSNN: High energy efficiency spike-train level spiking neural networks with spatio-temporal conversion. Neurocomputing, 607: 128364

  20. [28]

    Zhang, Y.; Inoue, K.; Nakajima, M.; Hashimoto, T.; Kuniyoshi, Y.; and Nakajima, K. 2024. Training Spiking Neural Networks via Augmented Direct Feedback Alignment. arXiv preprint arXiv:2409.07776

  21. [29]

    Zheng, H.; Wu, Y.; Deng, L.; Hu, Y.; and Li, G. 2021. Going deeper with directly-trained larger spiking neural networks. In Proceedings of the AAAI conference on artificial intelligence, volume 35, 11062--11070

  22. [30]

    , " * write output.state after.block = add.period write newline

    ENTRY address archivePrefix author booktitle chapter edition editor eid eprint howpublished institution isbn journal key month note number organization pages publisher school series title type volume year label extra.label sort.label short.list INTEGERS output.state before.all...

  23. [31]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 gl...

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.