Pith. sign in

REVIEW 5 major objections 6 minor 104 references

Spiking Neural Predictive Coding for Continual Learning from Data Streams

T0 review · 5 major / 6 minor · reviewed 2026-08-14 · deepseek-v4-flash

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

Pith's one-line read This paper claims that a spiking predictive-coding network with purely local spike-triggered weight updates can learn online from one-pass data streams, matching multi-epoch spiking networks on MNIST and forgetting less on sequential…

desk verdict A spiking predictive-coding network with a promising local rule, but the central equations are inconsistent and the empirics are hard to verify. read the letter →

arxiv 1908.08655 v3 pith:T4QNAEVB submitted 2019-08-23 cs.NE cs.LGq-bio.NC

Alexander Ororbia This is my paper · ORCID
classification cs.NEcs.LGq-bio.NC
keywords spikingneuralnetworkspredictivecodingcontinuallearningonlinesemi-supervisedlocalrepresentationalignmentleakyintegrate-and-firecatastrophicforgetting
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 sets out to establish that the iterative guess-and-check loop of neural predictive coding can be rebuilt with spiking neurons and still learn useful representations from one-pass data streams, without backpropagation or even spike-timing-dependent plasticity. The model predicts each layer's filtered spike activity, computes mismatch signals in dedicated error neurons, and adjusts synapses by a local outer-product rule whenever a spike arrives. If correct, spiking networks gain an online, biologically plausible learning rule that suits neuromorphic hardware, and the sparse spike codes appear to reduce catastrophic forgetting on sequential tasks. The reported evidence includes $4.72\%$ MNIST test error after a single pass, better average accuracy than its baselines on Split MNIST and NotMNIST, and usable semi-supervised learning as labels become rare.

What carries the argument

The mechanism that carries the argument is the spiking predictive-coding loop closed by dedicated error neurons and a spike-triggered local update rule. Predictions $z^\mu_\ell = W_\ell s_\ell(t)$ are compared against low-pass filtered spike traces $z_\ell(t)$ to produce error signals $e_\ell(t) = z^\mu_\ell - z_\ell(t)$; these errors are fed back through error synapses $E_\ell$ into the recurrent current equation $J_\ell(t) = (1-\kappa)J_\ell(t) + \kappa(-\gamma_J J_\ell(t) + \varphi(-e_\ell(t) + E_\ell e_{\ell-1}(t)))$ for intermediate layers, with only $E_L e_{L-1}(t)$ for the top layer. Learning is by Spike-Triggered Local Representation Alignment (ST-LRA), the event-driven rule $\Delta W_\ell = e_{\ell-1}(t) s_\ell(t)^\top$, $\Delta E_\ell = -\beta s_\ell(t) e_{\ell-1}(t)^\top$, an error-driven Hebbian outer-product update analogous to the delta rule. The paper uses a leaky integrate-and-fire spike-response model, but states the same three computations would accommodate richer neuron models such as Izhikevich or Hodgkin-Huxley units.

What would settle it

Train the reported four-layer SpNCN on MNIST with the error signals $e_\ell(t)$ replaced by independent noise of the same magnitude during learning, so that ST-LRA still updates but carries no predictive information: if test error stays near $4.72\%$, the mismatch computation is not the source of learning, and if it collapses toward chance, the predictive-coding error is confirmed as the mechanism.

Watch

Extended reading notes

Core claim

The central discovery claimed is that a spiking network built on predictive coding can learn online without a global error signal or repeated exposure to data. At every simulated time step, a layer $\ell$ predicts the filtered spike trace of another population, $z^\mu_\ell = W_\ell s_\ell(t)$, error neurons form the mismatch $e_\ell(t) = z^\mu_\ell - z_\ell(t)$, and these error activities are routed through error synapses $E_\ell$ back into the membrane currents $J_\ell(t)$ that drive the spike-response model. Synaptic change happens only when spikes occur, under the Spike-Triggered Local Representation Alignment rule $\Delta W_\ell = e_{\ell-1}(t) s_\ell(t)^\top$ and $\Delta E_\ell = -\beta s_\ell(t) e_{\ell-1}(t)^\top$. The paper reports that with four layers of leaky integrate-and-fire units, this loop reaches $4.72\%$ MNIST test error in one online pass, predicts a bouncing-ball video stream with lower error than a frame-repetition baseline, keeps working when most labels are missing, and retains more accuracy across Split MNIST and NotMNIST task streams than an equivalently sized backprop-trained ANN or a spiking network trained with derivative-free broadcast feedback alignment.

Load-bearing premise

The whole approach depends on the assumption that the mismatch signals computed at each layer, together with the local spike-triggered weight updates, are enough to train every layer correctly, a premise the paper motivates by analogy to the delta rule but does not prove or isolate.

Editorial extensions

If this is right

  • A four-layer SpNCN reaches $4.72 \pm 0.11\%$ test error on MNIST after one online pass, placing it alongside spiking networks that train over many epochs and above the paper's implemented spiking baselines.
  • On the continual-learning streams Split MNIST and NotMNIST, the SpNCN achieves $76.455\%$ and $77.945\%$ average accuracy, outperforming the self-implemented spiking baseline and a backprop-trained ANN under task-boundary fuzzing.
  • With labels on only $0.5\%$ of stream samples, a two-layer SpNCN still gets $24.08\%$ MNIST error versus $37.35\%$ for the derivative-free broadcast feedback alignment baseline, showing the generative side makes unlabeled data usable.
  • On the bouncing-ball stream, the SpNCN's prequential squared error reached 6.672 versus 10.225 for a frame-repetition baseline, and it made only about 46,393 lower-layer and 28,998 upper-layer weight updates over 300,000 simulation steps.
  • Because ST-LRA is local and event-driven, the paper expects it to combine with STDP and to fit neuromorphic hardware where sparse spike-driven updates translate to energy savings.

Reading between the lines

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

  • Inference: the claimed continual-learning gains are moderate, and the authors themselves note that sparsity alone will not solve catastrophic forgetting; a direct next test implied by their results is to combine ST-LRA with a complementary consolidation mechanism and measure forgetting on the same Split MNIST stream.
  • Inference: the generative side of the model suggests the same spike-driven loop could be used for sequence prediction and motor control, not just classification, by treating prediction error on future sensor frames as the learning signal in an event-camera or robotics stream.
  • Inference: because the paper does not compare its spiking version against a rate-coded version with the same local rule, a useful test is to run the same architecture with continuous units and non-spiking local representation alignment on the one-pass benchmarks to see whether spiking itself, rather than the update rule, drives the reported robustness.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

5 major / 6 minor

Summary. The paper proposes the Spiking Neural Coding Network (SpNCN), a recurrent network of leaky integrate-and-fire neurons trained by a spike-triggered local representation alignment (ST-LRA) update derived from predictive coding. The model is evaluated online and in continuous time on a bouncing-ball prediction task, on MNIST, Fashion-MNIST, Stanford OCR, and Caltech-101 classification, on semi-supervised variants of MNIST/Fashion-MNIST, and on Split-MNIST/NotMNIST continual-learning benchmarks. The central claims are that ST-LRA provides a local, backprop-free credit-assignment rule for multi-layer spiking networks, that SpNCN is competitive with existing SNNs after a single pass over the data, that it can exploit unlabeled data, and that its sparse spike-based representations forget less when tasks are presented sequentially.

Significance. If the learning mechanism performs as claimed, the paper offers a genuinely local, online, spike-based alternative to backpropagation-style training for spiking networks and a concrete test of the hypothesis that sparsity mitigates catastrophic forgetting. The paper has real strengths: all classification results are evaluated on held-out test sets rather than training data; the authors implement and compare against derivative-free spiking baselines under the same LIF SRM; the sparse weight-update counts in the bouncing-ball experiment give a concrete computational-economy datum; and the semi-supervised comparison is an informative stress test. The significance is conditional, however, because the central learning rule is not consistently specified and key simulation hyperparameters are missing.

major comments (5)
  1. [Section 2.2, Eqs. (2) and (5), Algorithm 1] The learning rule as printed is not internally consistent. Eq. (2) defines e_l(t) = W_l s_l(t) − z_l(t), so W_l appears to predict layer l's own filtered activity from its own spikes, whereas Eq. (5) updates W_l with ΔW_l = e_{l−1}(t)(s_l(t))^T, which is a predictive-coding update only under an off-by-one indexing convention in which W_l predicts z_{l−1}(t) from s_l(t). The subsequent sentence "W_l ← W_l − α_u ΔE_l and E_l ← E_l − α_u ΔW_l" then swaps the two matrices relative to their definitions, and the shapes of W_l and E_l are never stated. Algorithm 1 repeats Eqs. (2) and (5) without resolving the indexing. Since no code is provided, a reader cannot determine which variant produced Tables 1–4; if the equations are implemented literally, the rule may not be performing predictive-coding credit assignment at all. Please restate the model with consistent subscripts, specify the dimensions of all weight matrices, and reconcile Algorithm 1 with the equations.
  2. [Sections 2.2 and 3.2] The firing threshold v_thr is never reported. The text states only that a threshold is chosen and that voltages operate in the [0,1] decivolt range, while the MNIST experiments report 4 layers of 1000 LIF units, T_st = 100 ms, Δt = 0.25 ms, α = 0.0025, β = 1.0, and K = 63.75 Hz, but not the threshold, the membrane time constant τ_m, R_m, γ_m, τ_f, or whether an inter-stimulus interval T_ist was used. Because the threshold controls spiking frequency, the sparse spike-triggered updates, and hence the online learning dynamics, the headline 4.72% single-pass error cannot be reproduced or checked against the claim that sparse activity drives learning. Please provide a complete hyperparameter table for every experiment, including the bouncing-ball and continual-learning settings.
  3. [Table 4] The reported standard errors (±0.001% to ±0.003% over 10 trials) are implausibly small. For a binary accuracy near 0.76 on a held-out test set of even 10,000 examples, the per-trial standard deviation is roughly 0.4 percentage points and the standard error of the mean is about 0.13 percentage points, so the printed error bars are one to two orders of magnitude too small. As written, the error bars imply either that the metric is not what is stated, a typo, or an averaging artifact. This matters because the continual-learning advantage over the SNN baseline is the main quantitative support for the "less forgetting" claim, and its precision must be reported correctly.
  4. [Section 3.1, Figure 3] The bouncing-ball result is a single run: one stream of K = 2000 frames, one simulated 30-second sequence, and no repeated seeds, initializations, or error bars, and the 1000-frame train/freeze boundary is not justified. The Frame(t−1) comparison is useful, but a single-run pSE difference (6.672 vs 10.225) is anecdotal evidence for the model's predictive-tracking ability, which is one of the paper's three experimental pillars. Please report results over multiple random streams and initializations, with variance or interquartile ranges.
  5. [Table 4 and Section 3 ("On Catastrophic Forgetting")] The continual-learning experiments report only aggregate average accuracy (ACC) after the full stream. Aggregate accuracy does not directly measure forgetting; a model can achieve high ACC while still overwriting earlier tasks if later tasks are easier or if the final average is dominated by later performance. Since the stated goal is to determine whether spiking sparsity reduces forgetting, please also report per-task accuracy after each task, backward transfer, or an explicit forgetting measure such as final average per-task accuracy versus peak per-task accuracy.
minor comments (6)
  1. [Equation (1)] The first form of the trace filter writes z_l(t) on both sides; please use an explicit time index such as z_l(t+Δt) to remove the ambiguity between an in-place recurrence and a fixed-point equation.
  2. [Table 3] The caption says the columns vary "the proportion of samples that arrive labeled," but the first column is 0% and is described as fully supervised; please clarify whether the column percentages refer to labeled or unlabeled samples.
  3. [Table 4 and Section 3.1] Table 4 labels the baseline "SNN df-BDA," which should read "SNN df-BFA," and Section 3.1 plus Figure 3 use "SpTNCN" where "SpNCN" is meant.
  4. [Section 3.2, Table 1 caption] The caption says performance was "averaged over 10 trails"; this should be "10 trials."
  5. [Appendix (df-BFA and df-DRTP)] No hyperparameters are reported for the df-BFA and df-DRTP baselines (learning rates, numbers of layers and units, feedback weight initialization, training epochs, tuning procedure). Please provide these settings so the comparisons in Tables 1–3 cannot be attributed to under-tuned baselines.
  6. [Algorithm 1] In COMPUTE STATES, currents and voltages are updated layer by layer before the prediction/error loop; please state explicitly whether e_l(t) is computed from the post-update z_l(t) or from the previous value, since this timing changes the effective ST-LRA update.

Circularity Check

0 steps flagged · score 1.0 of 10

No significant circularity: the paper's empirical claims rest on external benchmark evaluations rather than fitted constants or self-citation chains.

full rationale

The paper's core claims—classification accuracy, online semi-supervised performance, reduced forgetting, and computational economy—are supported by measurements on held-out external benchmarks: MNIST test error in Table 1, Fashion MNIST / Stanford OCR / Caltech 101 in Table 2, semi-supervised errors in Table 3, and Split MNIST / NotMNIST ACC in Table 4. These quantities are not recovered from fitted constants or from the model equations by construction; they are empirical outcomes of running the described network on data. The ST-LRA learning rule (Eq. 5) is adopted from the author's prior LRA/NPC line of work ([72,73,74,75,77]), and self-citations appear in the motivation, but the present paper does not ask the reader to accept its results on the authority of those citations: it supplies its own spiking implementation and direct comparisons against df-BFA and df-DRTP baselines. No uniqueness theorem, no imported ansatz, and no definitional identity is used to force the main results. A separate concern about the consistency of Eq. 2 and Eq. 5, and about the weight-application line swapping ΔW and ΔE, is a correctness/reproducibility defect, not a circularity, because the published equations do not define the target accuracy in terms of the update's own outputs. Accordingly, the finding is 'no significant circularity,' with score 1 reflecting only the presence of non-load-bearing self-citations.

Assumptions & free parameters 7 free parameters · 4 assumptions · 0 invented entities

The model relies on standard LIF dynamics and Poisson encoding rather than new physical entities. The learning rule ST-LRA is a spiking adaptation of the author's earlier local representation alignment; the main unproven element is the sufficiency of local error-based updates for deep credit assignment in spiking networks.

free parameters (7)
  • Step size alpha_u = 0.0025
    Learning rate for ST-LRA weight updates, chosen by hand; no schedule or sensitivity analysis reported (Section 3.2).
  • Error weight coefficient beta = 1.0 (0.9 suggested earlier)
    Controls speed of error synapse evolution in Eq. 5; set to 1.0 in experiments.
  • Maximum input firing rate K = 63.75 Hz
    Scales Poisson spike rates for image inputs; chosen to match Diehl and Cook [19].
  • Spike threshold v_thr = not reported numerically
    Determines when LIF neurons emit spikes; only mentioned that voltages operate in [0,1] decivolt range (Section 2.2).
  • Weight column norm bound = 20
    Maximum Euclidean norm of weight matrix columns to prevent explosion; hand-set (Section 2.2).
  • Stimulus presentation time T_st = 100 ms for image tasks; 30 ms for bouncing ball
    Duration each input is presented; affects number of spike samples per pattern.
  • Integration step Delta t = 0.25 ms (0.1 ms for bouncing ball)
    Simulation time step for Euler integration; chosen by the author.
assumptions (4)
  • domain assumption Leaky integrate-and-fire (LIF) neuron model adequately captures computation for pattern recognition.
    The paper builds the SpNCN on LIF units and argues the framework extends to Hodgkin-Huxley, but all experiments use LIF.
  • domain assumption Poisson spike encoding faithfully represents input patterns.
    Input images are converted to Poisson spike trains via rate scaling; the paper does not compare other encoding schemes.
  • ad hoc to paper Error neurons and local representation alignment provide correct credit assignment in spiking networks.
    ST-LRA is adopted from prior non-spiking work; no proof or analysis is given that these local updates train hidden layers in the spiking recurrent setting, only empirical results.
  • domain assumption Single presentation of each stream sample is sufficient for learning.
    The stream setting processes each example once; the model must extract enough information from T_st milliseconds of spikes.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Spiking Neural Predictive Coding for Continual Learning from Data Streams." pith.science (2026). https://pith.science/paper/T4QNAEVB

@misc{pith2026190808655,
  author       = {Pith},
  title        = {Pith review of: Spiking Neural Predictive Coding for Continual Learning from Data Streams},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/T4QNAEVB}},
  note         = {Machine review of arXiv:1908.08655}
}
read the original abstract

For energy-efficient computation in specialized neuromorphic hardware, we present spiking neural coding, an instantiation of a family of artificial neural models grounded in the theory of predictive coding. This model, the first of its kind, works by operating in a never-ending process of "guess-and-check", where neurons predict the activity values of one another and then adjust their own activities to make better future predictions. The interactive, iterative nature of our system fits well into the continuous time formulation of sensory stream prediction and, as we show, the model's structure yields a local synaptic update rule, which can be used to complement or as an alternative to online spike-timing dependent plasticity. In this article, we experiment with an instantiation of our model consisting of leaky integrate-and-fire units. However, the framework within which our system is situated can naturally incorporate more complex neurons such as the Hodgkin-Huxley model. Our experimental results in pattern recognition demonstrate the potential of the model when binary spike trains are the primary paradigm for inter-neuron communication. Notably, spiking neural coding is competitive in terms of classification performance and experiences less forgetting when learning from task sequence, offering a more computationally economical, biologically-plausible alternative to popular artificial neural networks.

Figures

Figures reproduced from arXiv: 1908.08655 by the authors.

Figure 1
Figure 1. A 2-layer spiking neural coding network architecture. Green diamonds indicate error units (e 0 (t), e 1 (t)), which, in this model, compute the amount of mismatch between the current predictions (z 0 µ, z 1 µ) and the target signal traces (z 0 (t), z 1 (t)). Variables s 0 (t), s 1 (t) represent the binary spikes (or output of the underlying spike response model) of a particular vector grouping of neuronal units at t… view at source ↗
Figure 2
Figure 2. A demonstration of a spike train generated by a single LIF neuron within the SpNCN over time. Bottom diagram shows the [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗
Figure 3
Figure 3. SpNCN tracking performance on the bouncing ball [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: t-SNE visualizations of a fully supervised SpNCN (left) and an unsupervised SpNCN (right) (for MNIST). [PITH_FULL_IMAGE:figures/full_fig_p009_4.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

104 extracted references · 75 canonical work pages

  1. [1]

    Lateral competition for cortical space by layer-specific horizontal circuits

    ADESNIK , H., AND SCANZIANI , M. Lateral competition for cortical space by layer-specific horizontal circuits. Nature 464, 7292 (2010), 1155

    2010

  2. [2]

    Y., AND DAYAN, P

    A NGELA , J. Y., AND DAYAN, P. Uncertainty, neuromodulation, and attention. Neuron 46, 4 (2005), 681–692

    2005

  3. [3]

    Predictive coding of speech signals and subjective error criteria

    ATAL, B., AND SCHROEDER , M. Predictive coding of speech signals and subjective error criteria. IEEE Transactions on Acoustics, Speech, and Signal Processing 27, 3 (1979), 247–254

    1979

  4. [4]

    M., U SREY, W

    BASTOS , A. M., U SREY, W. M., A DAMS , R. A., M ANGUN , G. R., F RIES , P., AND FRISTON , K. J. Canonical microcircuits for predictive coding. Neuron 76, 4 (2012), 695–711

    2012

  5. [5]

    Simultaneous unsupervised and supervised learning of cognitive functions in biologically plausible spiking neural networks

    BEKOLAY, T., K OLBECK , C., AND ELIASMITH , C. Simultaneous unsupervised and supervised learning of cognitive functions in biologically plausible spiking neural networks. In Proceedings of the Annual Meeting of the Cognitive Science Society (2013), vol. 35

    2013

  6. [6]

    H., S UMMERFIELD , C., M ORIN , E

    BELL , A. H., S UMMERFIELD , C., M ORIN , E. L., M ALECEK , N. J., AND UNGERLEIDER , L. G. Encoding of stimulus probability in macaque inferior temporal cortex. Current Biology 26, 17 (2016), 2280–2290

    2016

  7. [7]

    D., AND KRICHMAR , J

    BEYELER , M., D UTT, N. D., AND KRICHMAR , J. L. Categorization and decision-making in a neurobiologically plausible spiking network using a stdp-like learning rule. Neural Networks 48 (2013), 109–124

    2013

  8. [8]

    Synaptic modifications in cultured hippocampal neurons: dependence on spike timing, synaptic strength, and postsynaptic cell type

    BI, G.- Q., AND POO, M.- M. Synaptic modifications in cultured hippocampal neurons: dependence on spike timing, synaptic strength, and postsynaptic cell type. Journal of neuroscience 18, 24 (1998), 10464–10472

    1998

Show all 104 references
  1. [9]

    Synaptic modification by correlated activity: Hebb’s postulate revisited

    BI, G.- Q., AND POO, M.- M. Synaptic modification by correlated activity: Hebb’s postulate revisited. Annual review of neuroscience 24, 1 (2001), 139–166

    2001

  2. [10]

    Neuromorphic microchips

    B OAHEN , K. Neuromorphic microchips. Scientific American 292, 5 (2005), 56–63

    2005

  3. [11]

    K., AND DENÈVE , S

    BOERLIN , M., M ACHENS , C. K., AND DENÈVE , S. Predictive coding of dynamical variables in balanced spiking networks. PLoS computational biology 9, 11 (2013), e1003258

    2013

  4. [12]

    M., K OK, J

    BOHTE , S. M., K OK, J. N., AND LA POUTRE , H. Error-backpropagation in temporally encoded networks of spiking neurons. Neurocomputing 48, 1-4 (2002), 17–37

    2002

  5. [13]

    M., S ENN , W., AND FUSI , S

    BRADER , J. M., S ENN , W., AND FUSI , S. Learning real-world stimuli in a neural network with spike-driven synaptic dynamics. Neural computation 19, 11 (2007), 2881–2912

    2007

  6. [14]

    W., AND AGGLETON , J

    BROWN , M. W., AND AGGLETON , J. P. Recognition memory: what are the roles of the perirhinal cortex and hippocampus? Nature Reviews Neuroscience 2, 1 (2001), 51

    2001

  7. [15]

    Pattern recogni- tion with spiking neural networks: a simple training method

    CHRISTOPHE , F., M IKKONEN , T., A NDALIBI , V., KOSKIMIES , K., AND LAUKKARINEN , T. Pattern recogni- tion with spiking neural networks: a simple training method. In SPLST (2015), pp. 296–308

    2015

  8. [16]

    H., D IMOU , G., J OSHI , P., IMAM , N., J AIN , S., ET AL

    DAVIES , M., S RINIVASA , N., L IN, T.-H., C HINYA , G., C AO, Y., C HODAY, S. H., D IMOU , G., J OSHI , P., IMAM , N., J AIN , S., ET AL . Loihi: A neuromorphic manycore processor with on-chip learning. IEEE Micro 38, 1 (2018), 82–99

    2018

  9. [17]

    Correlation between neural spike trains increases with firing rate

    DE LA ROCHA , J., D OIRON , B., S HEA -B ROWN , E., J OSI ´C, K., AND REYES , A. Correlation between neural spike trains increases with firing rate. Nature 448, 7155 (2007), 802

    2007

  10. [18]

    Bert: Pre-training of deep bidirectional transformers for language understanding

    DEVLIN , J., C HANG , M.-W., L EE, K., AND TOUTANOVA , K. Bert: Pre-training of deep bidirectional transformers for language understanding. arXiv preprint arXiv:1810.04805 (2018)

    2018 arXiv

  11. [19]

    U., AND COOK , M

    DIEHL , P. U., AND COOK , M. Unsupervised learning of digit recognition using spike-timing-dependent plasticity. Frontiers in computational neuroscience 9(2015), 99

    2015

  12. [20]

    U., N EIL , D., B INAS , J., C OOK , M., L IU, S.-C., AND PFEIFFER , M

    DIEHL , P. U., N EIL , D., B INAS , J., C OOK , M., L IU, S.-C., AND PFEIFFER , M. Fast-classifying, high-accuracy spiking deep networks through weight and threshold balancing. In 2015 International Joint Conference on Neural Networks (IJCNN) (2015), IEEE, pp. 1–8. 12 Preprint...

    2015

  13. [21]

    U., Z ARRELLA , G., C ASSIDY , A., P EDRONI , B

    DIEHL , P. U., Z ARRELLA , G., C ASSIDY , A., P EDRONI , B. U., AND NEFTCI , E. Conversion of artificial recurrent neural networks to spiking neural networks for low-power neuromorphic hardware. In 2016 IEEE International Conference on Rebooting Computing (ICRC) (2016), IEEE, pp. 1–8

    2016

  14. [22]

    Metalearning and neuromodulation

    D OYA, K. Metalearning and neuromodulation. Neural Networks 15, 4-6 (2002), 495–506

    2002

  15. [23]

    Semi-supervised classification with hybrid genera- tive/discriminative methods

    DRUCK , G., P AL, C., M CCALLUM , A., AND ZHU, X. Semi-supervised classification with hybrid genera- tive/discriminative methods. In Proceedings of the 13th ACM SIGKDD international conference on Knowledge discovery and data mining (2007), ACM, pp. 280–289

    2007

  16. [24]

    ELIASMITH , C., AND ANDERSON , C. H. Neural engineering: Computation, representation, and dynamics in neurobiological systems. MIT press, 2004

    2004

  17. [25]

    C., C HOO , X., B EKOLAY, T., D EWOLF, T., TANG , Y., AND RASMUSSEN , D

    ELIASMITH , C., S TEWART, T. C., C HOO , X., B EKOLAY, T., D EWOLF, T., TANG , Y., AND RASMUSSEN , D. A large-scale model of the functioning brain. science 338, 6111 (2012), 1202–1205

    2012

  18. [26]

    K., P ETERSEN , A

    FISER , A., M AHRINGER , D., O YIBO , H. K., P ETERSEN , A. V., L EINWEBER , M., AND KELLER , G. B. Experience-dependent spatial expectations in mouse visual cortex. Nature neuroscience 19, 12 (2016), 1658

    2016

  19. [27]

    FRENCH , R. M. Catastrophic forgetting in connectionist networks. Trends in cognitive sciences 3, 4 (1999), 128–135

    1999

  20. [28]

    Learning without feedback: Direct random target projection as a feedback-alignment algorithm with layerwise feedforward training

    FRENKEL , C., L EFEBVRE , M., AND BOL, D. Learning without feedback: Direct random target projection as a feedback-alignment algorithm with layerwise feedforward training. arXiv preprint arXiv:1909.01311 (2019)

    2019 arXiv

  21. [29]

    Predictive coding under the free-energy principle

    FRISTON , K., AND KIEBEL , S. Predictive coding under the free-energy principle. Philosophical Transactions of the Royal Society B: Biological Sciences 364, 1521 (2009), 1211–1221

    2009

  22. [30]

    B., G ALLUPPI , F., T EMPLE , S., AND PLANA , L

    FURBER , S. B., G ALLUPPI , F., T EMPLE , S., AND PLANA , L. A. The spinnaker project. Proceedings of the IEEE 102, 5 (2014), 652–665

    2014

  23. [31]

    GAMA , J., S EBASTIÃO , R., AND RODRIGUES , P. P. On evaluating stream learning algorithms. Machine learning 90, 3 (2013), 317–346

    2013

  24. [32]

    Understanding the difficulty of training deep feedforward neural networks

    GLOROT, X., AND BENGIO , Y. Understanding the difficulty of training deep feedforward neural networks. In Proceedings of the thirteenth international conference on artificial intelligence and statistics(2010), pp. 249–256

    2010

  25. [33]

    F., AND BRETTE , R

    G OODMAN , D. F., AND BRETTE , R. The brian simulator. Frontiers in neuroscience 3(2009), 26

    2009

  26. [34]

    Competitive learning: From interactive activation to adaptive resonance

    GROSSBERG , S. Competitive learning: From interactive activation to adaptive resonance. Cognitive science 11, 1 (1987), 23–63

    1987

  27. [35]

    Evolving spiking neural network controllers for autonomous robots

    HAGRAS , H., P OUNDS -CORNISH , A., C OLLEY , M., C ALLAGHAN , V., AND CLARKE , G. Evolving spiking neural network controllers for autonomous robots. InIEEE International Conference on Robotics and Automation,

  28. [36]

    A biologically plausible supervised learning method for spiking neural networks using the symmetric stdp rule

    HAO, Y., HUANG , X., D ONG , M., AND XU, B. A biologically plausible supervised learning method for spiking neural networks using the symmetric stdp rule. arXiv preprint arXiv:1812.06574 (2018)

    2018 arXiv

  29. [37]

    T., S IEGELMANN , H., AND KOZMA , R

    HAZAN , H., S AUNDERS , D., S ANGHAVI , D. T., S IEGELMANN , H., AND KOZMA , R. Unsupervised learning with self-organizing spiking neural networks. In 2018 International Joint Conference on Neural Networks (IJCNN) (2018), IEEE, pp. 1–6

    2018

  30. [38]

    Identity mappings in deep residual networks

    HE, K., Z HANG , X., R EN, S., AND SUN, J. Identity mappings in deep residual networks. In European conference on computer vision (2016), Springer, pp. 630–645

    2016

  31. [39]

    O., ET AL

    H EBB , D. O., ET AL . The organization of behavior, 1949

    1949

  32. [40]

    Deep neural networks for acoustic modeling in speech recognition

    HINTON , G., D ENG , L., Y U, D., D AHL , G., M OHAMED , A.- R., JAITLY, N., S ENIOR , A., V ANHOUCKE , V., NGUYEN , P., K INGSBURY , B., ET AL . Deep neural networks for acoustic modeling in speech recognition. IEEE Signal processing magazine 29 (2012)

    2012

  33. [41]

    E., AND MCCLELLAND , J

    HINTON , G. E., AND MCCLELLAND , J. L. Learning representations by recirculation. In Neural information processing systems (1988), pp. 358–366

    1988

  34. [42]

    L., AND HUXLEY , A

    HODGKIN , A. L., AND HUXLEY , A. F. A quantitative description of membrane current and its application to conduction and excitation in nerve. The Journal of physiology 117, 4 (1952), 500–544

    1952

  35. [43]

    Improved margin multi-class classification using dendritic neurons with morphological learning

    HUSSAIN , S., L IU, S.-C., AND BASU , A. Improved margin multi-class classification using dendritic neurons with morphological learning. In 2014 IEEE International Symposium on Circuits and Systems (ISCAS) (2014), IEEE, pp. 2640–2643

    2014

  36. [44]

    A self-driving robot using deep convolutional neural networks on neuromorphic hardware

    HWU, T., I SBELL , J., O ROS, N., AND KRICHMAR , J. A self-driving robot using deep convolutional neural networks on neuromorphic hardware. In 2017 International Joint Conference on Neural Networks (IJCNN) (2017), IEEE, pp. 635–641. 13 Preprint, Work in Progress

    2017

  37. [45]

    Frontiers in neuromorphic engineering

    INDIVERI , G., AND HORIUCHI , T. Frontiers in neuromorphic engineering. Frontiers in Neuroscience 5(2011), 118

    2011

  38. [46]

    IZHIKEVICH , E. M. Simple model of spiking neurons. IEEE Transactions on neural networks 14, 6 (2003), 1569–1572

    2003

  39. [47]

    M., O SINDERO , S., V INYALS , O., G RAVES , A., S ILVER , D., AND KAVUKCUOGLU , K

    JADERBERG , M., C ZARNECKI , W. M., O SINDERO , S., V INYALS , O., G RAVES , A., S ILVER , D., AND KAVUKCUOGLU , K. Decoupled neural interfaces using synthetic gradients. In Proceedings of the 34th International Conference on Machine Learning-Volume 70(2017), JMLR. org, pp. 1627–1635

    2017

  40. [48]

    Improving protein fold recognition by deep learning networks

    JO, T., H OU, J., E ICKHOLT , J., AND CHENG , J. Improving protein fold recognition by deep learning networks. Scientific reports 5 (2015), 17573

    2015

  41. [49]

    R., G ANJTABESH , M., T HORPE , S

    KHERADPISHEH , S. R., G ANJTABESH , M., T HORPE , S. J., AND MASQUELIER , T. Stdp-based spiking deep convolutional neural networks for object recognition. Neural Networks 99 (2018), 56–67

    2018

  42. [50]

    Retrospective and prospective coding for predicted reward in the sensory thalamus

    KOMURA , Y., TAMURA , R., U WANO, T., N ISHIJO , H., K AGA, K., AND ONO, T. Retrospective and prospective coding for predicted reward in the sensory thalamus. Nature 412, 6846 (2001), 546

    2001

  43. [51]

    KRIZHEVSKY , A., S UTSKEVER , I., AND HINTON , G. E. Imagenet classification with deep convolutional neural networks. In Advances in neural information processing systems (2012), pp. 1097–1105

    2012

  44. [52]

    KROTOV, D., AND HOPFIELD , J. J. Unsupervised learning by competing hidden units. Proceedings of the National Academy of Sciences 116, 16 (2019), 7723–7731

    2019

  45. [53]

    Learning algorithms for the classification restricted boltzmann machine

    LAROCHELLE , H., M ANDEL , M., P ASCANU , R., AND BENGIO , Y. Learning algorithms for the classification restricted boltzmann machine. Journal of Machine Learning Research 13, Mar (2012), 643–669

    2012

  46. [54]

    H., D ELBRUCK , T., AND PFEIFFER , M

    LEE, J. H., D ELBRUCK , T., AND PFEIFFER , M. Training deep spiking neural networks using backpropagation. Frontiers in neuroscience 10(2016), 508

    2016

  47. [55]

    On the relation between catastrophic interference and generalization in connectionist networks

    LEWANDOWSKY , S. On the relation between catastrophic interference and generalization in connectionist networks. Journal of Biological Systems 2, 03 (1994), 307–333

    1994

  48. [56]

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

    MAASS , W. Networks of spiking neurons: the third generation of neural network models. Neural networks 10, 9 (1997), 1659–1671

    1997

  49. [57]

    Fine-tuning and the stability of recurrent neural networks

    MACNEIL , D., AND ELIASMITH , C. Fine-tuning and the stability of recurrent neural networks. PLOS ONE 6, 9 (09 2011), 1–16

    2011

  50. [58]

    C., AND JOHNSTON , D

    MAGEE , J. C., AND JOHNSTON , D. A synaptically controlled, associative signal for hebbian plasticity in hippocampal neurons. Science 275, 5297 (1997), 209–213

    1997

  51. [59]

    MCCLOSKEY , M., AND COHEN , N. J. Catastrophic interference in connectionist networks: The sequential learning problem. The psychology of learning and motivation 24, 109 (1989), 92

    1989

  52. [60]

    Neurotransmitters as developmental signals

    MEIER , E., H ERTZ , L., AND SCHOUSBOE , A. Neurotransmitters as developmental signals. Neurochemistry international 19, 1-2 (1991), 1–15

    1991

  53. [61]

    MEROLLA , P., A RTHUR , J., A KOPYAN , F., I MAM , N., M ANOHAR , R., AND MODHA , D. S. A digital neurosynaptic core using embedded crossbar memory with 45pj per spike in 45nm. In 2011 IEEE custom integrated circuits conference (CICC)(2011), IEEE, pp. 1–4

    2011

  54. [62]

    A., A RTHUR , J

    MEROLLA , P. A., A RTHUR , J. V., A LVAREZ-I CAZA , R., C ASSIDY , A. S., S AWADA, J., A KOPYAN , F., JACKSON , B. L., I MAM , N., G UO, C., N AKAMURA , Y., ET AL . A million spiking-neuron integrated circuit with a scalable communication network and interface. Science 345, 61...

    2014

  55. [63]

    K., L I, L., AND DESIMONE , R

    MILLER , E. K., L I, L., AND DESIMONE , R. Activity of neurons in anterior inferior temporal cortex during a short-term memory task. Journal of Neuroscience 13, 4 (1993), 1460–1478

    1993

  56. [64]

    J., AND MASQUELIER , T

    MOZAFARI , M., G ANJTABESH , M., N OWZARI -DALINI , A., T HORPE , S. J., AND MASQUELIER , T. Bio- inspired digit recognition using reward-modulated spike-timing-dependent plasticity in deep convolutional networks. Pattern Recognition 94 (2019), 87–95

    2019

  57. [65]

    Event-driven contrastive divergence for spiking neuromorphic systems

    NEFTCI , E., D AS, S., P EDRONI , B., K REUTZ -D ELGADO , K., AND CAUWENBERGHS , G. Event-driven contrastive divergence for spiking neuromorphic systems. Frontiers in neuroscience 7 (2014), 272

    2014

  58. [66]

    Cognitive psychology: Classic edition

    N EISSER , U. Cognitive psychology: Classic edition. Psychology Press, 2014

    2014

  59. [67]

    Real-time classification and sensor fusion with a spiking deep belief network

    O’CONNOR , P., N EIL , D., L IU, S.-C., D ELBRUCK , T., AND PFEIFFER , M. Real-time classification and sensor fusion with a spiking deep belief network. Frontiers in neuroscience 7 (2013), 178

    2013

  60. [68]

    A., AND FIELD , D

    OLSHAUSEN , B. A., AND FIELD , D. J. Sparse coding with an overcomplete basis set: A strategy employed by v1? Vision research 37, 23 (1997), 3311–3325. 14 Preprint, Work in Progress

    1997

  61. [69]

    OORD , A. V. D., D IELEMAN , S., Z EN, H., S IMONYAN , K., V INYALS , O., G RAVES , A., K ALCHBRENNER , N., S ENIOR , A., AND KAVUKCUOGLU , K. Wavenet: A generative model for raw audio. arXiv preprint arXiv:1609.03499 (2016)

    2016 arXiv

  62. [70]

    O’REILLY, R. C. Six principles for biologically based computational models of cortical cognition. Trends in cognitive sciences 2, 11 (1998), 455–462

    1998

  63. [71]

    C., AND MUNAKATA , Y

    O’REILLY, R. C., AND MUNAKATA , Y. Computational explorations in cognitive neuroscience: Understanding the mind by simulating the brain. MIT press, 2000

    2000

  64. [72]

    L., AND KIFER , D

    ORORBIA , A., M ALI , A., G ILES , C. L., AND KIFER , D. Continual learning of recurrent neural networks by locally aligning distributed representations. arXiv preprint arXiv:1810.07411 (2018)

    2018 arXiv

  65. [73]

    ORORBIA , A., M ALI , A., K IFER , D., AND GILES , C. L. Lifelong neural predictive coding: Sparsity yields less forgetting when learning cumulatively. arXiv preprint arXiv:1905.10696 (2019)

    2019 arXiv

  66. [74]

    G., H AFFNER , P., R EITTER , D., AND GILES , C

    ORORBIA , A. G., H AFFNER , P., R EITTER , D., AND GILES , C. L. Learning to adapt by minimizing discrepancy. arXiv preprint arXiv:1711.11542 (2017)

    2017 arXiv

  67. [75]

    G., AND MALI , A

    ORORBIA , A. G., AND MALI , A. Biologically motivated algorithms for propagating local target representations. In Proceedings of the AAAI Conference on Artificial Intelligence (2019), vol. 33, pp. 4651–4658

    2019

  68. [76]

    G., M ALI , A., K ELLY, M

    ORORBIA , A. G., M ALI , A., K ELLY, M. A., AND REITTER , D. Like a Baby: Visually Situated Neural Language Acquisition. In Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics (Florence, Italy, 2019)

    2019

  69. [77]

    G., M ALI , A., K IFER , D., AND GILES , C

    ORORBIA , A. G., M ALI , A., K IFER , D., AND GILES , C. L. Deep credit assignment by aligning local representations. arXiv preprint arXiv:1803.01834 (2018)

    2018 arXiv

  70. [78]

    L., AND REITTER , D

    ORORBIA , I., A LEXANDER , G., G ILES , C. L., AND REITTER , D. Online semi-supervised learning with deep hybrid boltzmann machines and denoising autoencoders. arXiv preprint arXiv:1511.06964 (2015)

    2015 arXiv

  71. [79]

    L., AND REITTER , D

    ORORBIA II, A., G ILES , C. L., AND REITTER , D. Learning a deep hybrid model for semi-supervised text classification. In Proceedings of the 2015 Conference on Empirical Methods in Natural Language Processing (2015), pp. 471–481

    2015

  72. [80]

    G., M IKOLOV, T., AND REITTER , D

    ORORBIA II, A. G., M IKOLOV, T., AND REITTER , D. Learning simpler language models with the differential state framework. Neural Computation 29, 12 (2017), 3327–3352

    2017

  73. [81]

    G., R EITTER , D., W U, J., AND GILES , C

    ORORBIA II, A. G., R EITTER , D., W U, J., AND GILES , C. L. Online learning of deep hybrid architectures for semi-supervised categorization. In Machine Learning and Knowledge Discovery in Databases (Proceedings, ECML PKDD 2015), vol. 9284 of Lecture Notes in Computer Science....

    2015

  74. [82]

    Deep learning for regulatory genomics

    P ARK , Y., AND KELLIS , M. Deep learning for regulatory genomics. Nature biotechnology 33, 8 (2015), 825

    2015

  75. [83]

    On the difficulty of training recurrent neural networks

    PASCANU , R., M IKOLOV, T., AND BENGIO , Y. On the difficulty of training recurrent neural networks. In International conference on machine learning (2013), pp. 1310–1318

    2013

  76. [84]

    V., B LUMER , K., L IU, Y., M CCONNELL , M

    P OPLIN , R., V ARADARAJAN , A. V., B LUMER , K., L IU, Y., M CCONNELL , M. V., C ORRADO , G. S., P ENG , L., AND WEBSTER , D. R. Prediction of cardiovascular risk factors from retinal fundus photographs via deep learning. Nature Biomedical Engineering 2, 3 (2018), 158

    2018

  77. [85]

    Immunity to device variations in a spiking neural network with memristive nanodevices

    QUERLIOZ , D., B ICHLER , O., D OLLFUS , P., AND GAMRAT, C. Immunity to device variations in a spiking neural network with memristive nanodevices. IEEE Transactions on Nanotechnology 12, 3 (2013), 288–295

    2013

  78. [86]

    C., AND MILLER , E

    RAINER , G., R AO, S. C., AND MILLER , E. K. Prospective coding for objects in primate prefrontal cortex. Journal of Neuroscience 19, 13 (1999), 5493–5505

    1999

  79. [87]

    M., H AJAJ , N., H ARDT , M., L IU, P

    RAJKOMAR , A., O REN , E., C HEN , K., D AI, A. M., H AJAJ , N., H ARDT , M., L IU, P. J., L IU, X., M ARCUS , J., S UN, M., ET AL . Scalable and accurate deep learning with electronic health records. NPJ Digital Medicine 1, 1 (2018), 18

    2018

  80. [88]

    P., AND BALLARD , D

    RAO, R. P., AND BALLARD , D. H. Predictive coding in the visual cortex: a functional interpretation of some extra-classical receptive-field effects. Nature neuroscience 2, 1 (1999)

    1999

  81. [89]

    Connectionist models of recognition memory: constraints imposed by learning and forgetting functions

    RATCLIFF , R. Connectionist models of recognition memory: constraints imposed by learning and forgetting functions. Psychological review 97, 2 (1990), 285

    1990

  82. [90]

    Recurrent spiking networks solve planning tasks

    RUECKERT , E., K APPEL , D., T ANNEBERG , D., P ECEVSKI , D., AND PETERS , J. Recurrent spiking networks solve planning tasks. Scientific reports 6 (2016), 21142

    2016

  83. [91]

    P., AND TWEED , D

    SAMADI , A., L ILLICRAP , T. P., AND TWEED , D. B. Deep learning with dynamic spiking neurons and fixed feedback weights. Neural computation 29, 3 (2017), 578–602. 15 Preprint, Work in Progress

    2017

  84. [92]

    E., AND TAYLOR , G

    SUTSKEVER , I., H INTON , G. E., AND TAYLOR , G. W. The recurrent temporal restricted boltzmann machine. In Advances in Neural Information Processing Systems (2009), pp. 1601–1608

    2009

  85. [93]

    D., G REGOR , K., AND CUN, Y

    SZLAM , A. D., G REGOR , K., AND CUN, Y. L. Structured sparse coding via lateral inhibition. In Advances in Neural Information Processing Systems (2011), pp. 1116–1124

    2011

  86. [94]

    W., H INTON , G

    TAYLOR , G. W., H INTON , G. E., AND ROWEIS , S. T. Modeling human motion using binary latent variables. In Advances in neural information processing systems (2007), pp. 1345–1352

    2007

  87. [95]

    C., B ICHLER , O., AND DUPRET , A

    THIELE , J. C., B ICHLER , O., AND DUPRET , A. Event-based, timescale invariant unsupervised online deep learning with stdp. Frontiers in computational neuroscience 12(2018), 46

    2018

  88. [96]

    THRUN , S., AND MITCHELL , T. M. Lifelong robot learning. Robotics and autonomous systems 15, 1-2 (1995), 25–46

    1995

  89. [97]

    J., ET AL

    WERBOS , P. J., ET AL . Backpropagation through time: what it does and how to do it. Proceedings of the IEEE 78, 10 (1990), 1550–1560

    1990

  90. [98]

    WIDROW, B., AND HOFF, M. E. Adaptive switching circuits. Tech. rep., Stanford Univ Ca Stanford Electronics Labs, 1960

    1960

  91. [99]

    WISKOTT , L., AND SEJNOWSKI , T. J. Slow feature analysis: Unsupervised learning of invariances. Neural computation 14, 4 (2002), 715–770

    2002

  92. [100]

    Fashion-mnist: a novel image dataset for benchmarking machine learning algorithms

    X IAO, H., R ASUL , K., AND VOLLGRAF , R. Fashion-mnist: a novel image dataset for benchmarking machine learning algorithms. arXiv preprint arXiv:1708.07747 (2017)

    2017 arXiv

  93. [101]

    Continual learning of context-dependent processing in neural networks

    ZENG , G., C HEN , Y., C UI, B., AND YU, S. Continual learning of context-dependent processing in neural networks. Nature Machine Intelligence 1, 8 (2019), 364–372

    2019

  94. [102]

    Feedforward categorization on aer motion events using cortex-like features in a spiking neural network

    ZHAO, B., D ING , R., C HEN , S., L INARES -BARRANCO , B., AND TANG , H. Feedforward categorization on aer motion events using cortex-like features in a spiking neural network. IEEE transactions on neural networks and learning systems 26, 9 (2014), 1963–1978

    2014

  95. [103]

    O”) shape from an X-cross (“X

    ZMARZ , P., AND KELLER , G. B. Mismatch receptive fields in mouse visual cortex. Neuron 92, 4 (2016), 766–772. 16 Preprint, Work in Progress Appendix Algorithmic Specification In this section, we present the full pseudocode, i.e., Algorithm 1, for computing state updates and syn...

    2016

  96. [2004]

    Proceedings. ICRA’04. 2004(2004), vol. 5, IEEE, pp. 4620–4626

    2004

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.