Pith. sign in

REVIEW 4 major objections 7 minor 42 references

Maximizing Information in Neuron Populations for Neuromorphic Spike Encoding

T0 review · 4 major / 7 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read Adding neurons to a spike-encoding population, chosen greedily to maximize mutual information, raises both encoded information and classification accuracy, nearly matching task-specific tuning.

desk verdict A plausible, modest contribution — greedy MI-based tuning of LIF population thresholds — whose qualitative results are likely right, but the numbers rest on an extrapolated MI estimator with no fit diagnostics. read the letter →

arxiv 2412.08816 v2 pith:FXIFXXIB submitted 2024-12-11 cs.NE

Ahmad El Ferdaoussi , Eric Plourde , Jean Rouat This is my paper · ORCID
classification cs.NE
keywords spikeencodingpopulationcodingmutualinformationpartialdecompositionneuromorphicleakyintegrate-and-fireLIFneuronclassificationaccuracy
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

This paper tries to show that the encoding parameters of a population of spiking neurons do not need to be tuned for a specific task: greedily choosing each neuron's firing threshold to maximize the mutual information between the input signal and the resulting spike trains produces configurations whose classification accuracy is near the best achievable with task-specific tuning. The authors argue that population coding adds information beyond any single neuron—unique and synergistic contributions—and that these gains track accuracy as neurons are added. They demonstrate this in two unrelated tasks, blood-pressure pulse-wave classification and neural action-potential waveform classification, using leaky integrate-and-fire neurons. If the claim holds, mutual information can serve as a task-independent proxy for setting spike-encoding parameters, which matters because spike encoding is a bottleneck for neuromorphic systems and task-specific optimization can be expensive.

What carries the argument

The machinery is the recursive greedy maximization of joint mutual information. At each step the algorithm treats the population selected so far as one fixed random variable $W_1$, then searches over the firing threshold of a new leaky integrate-and-fire neuron to maximize $I(X; \{W_1, W_2\})$; the new neuron is accepted at the threshold giving the largest joint information, and the process repeats until a stopping condition. The estimated quantity is a discrete-continuous mutual information between signal amplitude and spike patterns, computed with a k-nearest-neighbors estimator and then extrapolated to an infinite observation window through the quadratic fit $I(X; W^T)=a+b/T+c/T^2$, whose intercept $a$ is taken as the 'true' mutual information.

What would settle it

Re-run the two protocols using a bias-corrected mutual-information estimator at a single sufficiently long window (or with bootstrap confidence intervals over many data subsamples) instead of the quadratic extrapolation, and compare the resulting parameter rankings and accuracy curves. If the alternative estimator changes which spike densities are selected, or if the extrapolated MI stops tracking held-out classification accuracy as the population grows, the central claim would be contradicted.

Watch

Extended reading notes

Core claim

On the paper's own terms, the central discovery is that a greedy, information-theoretic construction of a neuron population is both practically effective and nearly optimal. Starting with the neuron whose spike train maximizes $I(X; W_1)$, the algorithm fixes that neuron and chooses each next neuron to maximize the joint information $I(X; \{W_1, W_2\})$ between the stimulus and the whole population, with the previously chosen neurons treated as a single aggregate variable; this is inspired by the Partial Information Decomposition of joint information into redundant, unique, and synergistic atoms. In the pulse-wave task, tuning five such neurons raises the estimated mutual information from 3.058 to 6.116 bits and classification accuracy from 45.1% to 68.6%; in the action-potential task it raises MI from 1.876 to 3.187 bits and accuracy from 81.8% to 94.8%. The authors report that these accuracy values closely track the MI curve, match an exhaustive task-specific parameter search, and beat random parameter selection, with diminishing returns as neurons are added.

Load-bearing premise

The central claim stands or falls on the reliability of the estimated 'true' mutual information, which is obtained by extrapolating finite-window estimates to infinite window duration with a quadratic fit; if that extrapolation is biased or unstable, the greedy parameter choices, the reported MI gains, and the claimed link to accuracy would not be trustworthy.

Editorial extensions

If this is right

  • Mutual information can be used as a task-independent target when configuring spike encoders, removing the need to train or evaluate a downstream model during parameter search.
  • Population coding in neuromorphic encoders can yield joint gains beyond single-neuron information, with accuracy rising as neurons are added until diminishing returns set in.
  • The greedy algorithm can also serve as a principled way to choose how many neurons to include, by watching the MI curve flatten.
  • The same tuning recipe should transfer to other spike encoders and other signal types, since the objective does not depend on the classification model or task.

Reading between the lines

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

  • A natural extension the paper leaves implicit is to test whether MI-optimal parameters stay near-optimal when the classifier or feature extractor changes, since the current evidence uses one SVM with spike-count bins.
  • The paper does not report the redundant, unique, and synergistic atoms it invokes; computing those atoms directly would show whether the accuracy gains come mostly from unique or synergistic information, and could predict when adding neurons stops helping.
  • The quadratic extrapolation is a potential fragility; replacing it with modern bias-correction techniques could make the approach reliable on shorter recordings where the fit is less stable.
  • If the MI-accuracy link generalizes, the algorithm could serve as a model-free diagnostic for comparing different spike-encoding schemes before committing to a downstream neuromorphic pipeline.
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

4 major / 7 minor

Summary. The paper proposes a greedy algorithm for tuning the parameters of a population of leaky integrate-and-fire (LIF) neurons that encode a real-valued stimulus into spike trains. Starting from a single neuron whose firing threshold is chosen to maximize an estimate of mutual information I(X; W_1), the algorithm recursively adds neurons to maximize the joint mutual information I(X; {W_1, ..., W_m}), treating previously selected neurons as a fixed aggregate. The mutual information is estimated with a k-nearest-neighbor estimator and extrapolated to infinite observation-window duration via a quadratic fit in 1/T, with the intercept taken as the 'true' mutual information. The method is evaluated on two tasks: 13-way classification of blood pressure pulse waves (PWDB) and 21-way classification of simulated action potential waveforms (SYNTH Monotrode). The authors report that both mutual information and classification accuracy increase as neurons are added, that the mutual-information-selected thresholds achieve accuracy close to thresholds found by an exhaustive search for accuracy, and that they outperform random threshold selection. The central conclusion is that mutual information can serve as a task-independent proxy for tuning spike-encoding parameters in neuromorphic systems.

Significance. If the reported results are reliable, the paper offers a practical and conceptually appealing recipe for configuring neuromorphic encoding populations without task-specific labels, with potential computational savings. The use of two distinct datasets, a comparison against an exhaustive accuracy-oriented search, and a random-parameter baseline are strengths, and the qualitative consistency of the mutual-information/accuracy relationship across both applications is encouraging. However, the central quantitative claims rest on an unvalidated extrapolation of mutual-information estimates, and the accuracy comparisons lack uncertainty quantification; these issues currently prevent the stronger conclusions from being fully supported.

major comments (4)
  1. [Section 2.2, Eq. (5)] The definition of the 'true' mutual information as the intercept a of the quadratic fit I(X; W_T) = a + b/T + c/T^2 is not validated. The paper does not report the number of window durations T used, the range of T, the residuals of the fit, any goodness-of-fit statistic, or uncertainty estimates for a. Since every threshold selection in Figs. 5 and 7 and every mutual-information value in Tables 1 and 2 is derived from this extrapolated intercept, the claim that the algorithm maximizes true mutual information is not currently supported. Please provide fit diagnostics, a sensitivity analysis of the chosen T values, and error bars or bootstrap intervals for a; without these, the reported information gains and parameter choices may be artifacts of the extrapolation procedure.
  2. [Section 3.3 and Fig. 8] All classification accuracies are point estimates from a single 80/20 split, with no repeated splits, cross-validation, or confidence intervals. The claims that the proposed approach achieves 'near-optimal' accuracy and 'significantly outperforms' random parameter selection—the latter based on only five random trials—are therefore not statistically substantiated. Please report accuracy variability across multiple splits or bootstrap resamples and provide a larger number of random trials with a significance test to support the comparison.
  3. [Section 3.3 and Fig. 8] The 'exhaustive search' used to obtain the accuracy-optimal parameters is not specified: the grid resolution over spike densities, the data split used, and whether the same 260/1000 waveforms were used for parameter selection are all missing. Without these details, the near-optimality claim in Fig. 8 cannot be reproduced or evaluated. Please describe the exhaustive search procedure and, ideally, report the actual accuracy values and grid density.
  4. [Section 2.1 and Section 4] The algorithm is motivated by Partial Information Decomposition, and the abstract claims gains 'beyond what could be accounted for by each neuron separately,' but no PID atoms (redundant, unique, synergistic) are ever computed. The observed increase in joint mutual information over single-neuron mutual information does not by itself establish unique or synergistic contributions, because the individual mutual information of each added neuron is not reported and the greedy recursion fixes previous neurons. Either compute the PID decomposition (or at least report the single-neuron mutual information of each added neuron) or soften the synergy-related wording so that the claim matches what is actually measured.
minor comments (7)
  1. [Eq. (5)] The displayed equation for the quadratic extrapolation is garbled in the manuscript ('T b C'); it should read I(X; W_T) = a + b/T + c/T^2.
  2. [Section 2.1.2] In the sentence introducing the second neuron, 'J(X; W2)' should be 'I(X; W2)'.
  3. [Section 3.1.1] The database size is given as '437 4 virtual subjects'; this appears to be a typo for 4,374.
  4. [Section 3] The text says SVMs are implemented with 'sciki t-learn'; the correct name is scikit-learn.
  5. [Fig. 1 caption] The caption contains a garbled character sequence 'n:,;}' in the expression for the spike trains; please clean this up.
  6. [Section 3.1.2 and 3.2.2] The grid search over firing thresholds is described only as spanning the range of spike densities; the number of threshold values and the density resolution should be stated for reproducibility.
  7. [Section 3.3] The term 'near-optimal' is used without a quantitative definition; please specify a tolerance or margin relative to the exhaustive-search accuracy.

Circularity Check

1 steps flagged · score 2.0 of 10

The central accuracy claims are independently benchmarked; only the MI-gain/synergy statement is a restatement of the algorithm's objective.

  1. self definitional [Section 2.1.2 (Eqs. 1-3) and Section 3.1.3 (Application 1, m=2)]
    "The algorithm finds the encoding parameters of the second neuron by maximizing the joint information I(X; {W1, W2} ). Since W1 is fixed in this case and only W2 is variable, this effectively maximizes the information contributed uniquely by the second neuron, as well as the synergistic information that emerges in the population."

    From the paper's own PID identities (Eqs. 1-3), I(X;{W1,W2}) - I(X;W1) = Iunq2 + Isyn. With W1 fixed, maximizing the joint information is exactly maximizing this difference. Therefore the reported 'increase in encoded information ... contributed both uniquely by the second neuron and in synergy with the first neuron' is the same quantity the greedy step optimizes; it is a definitional restatement of the objective rather than an independent prediction. The classification-accuracy gains are measured independently, so the circularity is limited to the MI-gain/synergy wording.

full rationale

The paper's central empirical claim is that parameters tuned by maximizing estimated mutual information yield near-optimal classification accuracy. That claim is tested against an independent accuracy metric: SVM classification of held-out spike-count features, an exhaustive accuracy-based parameter search, and random parameter baselines. Accuracy is never used in the information-maximization loop, so the accuracy comparisons are not circular. The reported mutual-information gains, however, are partly self-referential: the greedy algorithm selects each added neuron by maximizing joint information, and since joint information is at least the previous neuron's individual information, the observed MI increase is guaranteed by the objective (and by information-theoretic monotonicity), not an empirical discovery. The further description of the gain as 'unique plus synergistic' follows directly from the PID definitions in Eqs. 1-3 and does not require an independent decomposition. A related robustness concern, distinct from circularity, is that the 'true' MI is defined as the intercept a of a quadratic extrapolation (Eqs. 5-6), making all reported MI values model outputs; the paper gives no fit diagnostics or error bars, which weakens the quantitative MI claims but does not make the accuracy validation circular. The self-citations [9] and [10] are background and are not load-bearing; the algorithm is fully described and its value rests on the independent accuracy experiments. Overall, no significant circularity; score 2 reflects the mild self-referential wording of the MI-gain and synergy statement.

Assumptions & free parameters 5 free parameters · 5 assumptions · 0 invented entities

The central claim rests on the reliability of the MI estimator and extrapolation, the chosen LIF model with its fixed decay, and the greedy selection heuristic. No new entities are introduced. The free parameters above are either hand-set or fitted to data as part of the MI estimation.

free parameters (5)
  • LIF decay constant delta = 0.5
    Set by hand in Section 2.3; controls the leak of the membrane potential and therefore the spike patterns. The paper does not optimize it or study its effect.
  • k in k-NN MI estimator = 3
    Recommended value from Ross 2014; a hyperparameter of the MI estimator in Section 2.2.
  • Quadratic extrapolation coefficients a, b, c = Not reported
    Fitted to MI estimates at several window durations in Section 2.2; a is reported as the 'true' MI. The values are not given, and no fit quality is shown.
  • Spike-count feature bins per train = 20 bins, 50% overlap
    Feature extraction choice in Section 3; affects classification accuracy but is not part of the MI optimization.
  • SVM RBF hyperparameters (C, gamma) = scikit-learn defaults
    The classifier uses fixed defaults (Section 3); these influence accuracy but are not tuned or reported.
assumptions (5)
  • standard math The Ross (2014) k-NN estimator provides unbiased mutual information estimates for mixed continuous/discrete variables with finite samples.
    Section 2.2 relies on this estimator; finite-sample bias is assumed handled by the extrapolation.
  • domain assumption The mutual information for window duration T follows I(X;W_T)=a+b/T+c/T^2, so the infinite-window MI is the intercept a.
    Section 2.2, Eq. (5); this asymptotic expansion is assumed to hold for LIF spike trains.
  • ad hoc to paper Greedy forward selection, fixing earlier neurons, reaches near-optimal population information and accuracy.
    Section 2.1.3 and Section 3.3; no comparison against non-greedy or joint optimization of all thresholds is provided.
  • domain assumption The instantaneous signal amplitude X is the random variable carrying task-relevant information.
    Section 2.2; classification labels may depend on waveform shape beyond instantaneous amplitude, but MI is computed on amplitude values.
  • domain assumption The LIF neuron with decay 0.5 is a representative encoder for generalizing the algorithm.
    Section 2.3 asserts generality but only tests one encoder model with one fixed decay.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Maximizing Information in Neuron Populations for Neuromorphic Spike Encoding." pith.science (2026). https://pith.science/paper/FXIFXXIB

@misc{pith2026241208816,
  author       = {Pith},
  title        = {Pith review of: Maximizing Information in Neuron Populations for Neuromorphic Spike Encoding},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/FXIFXXIB}},
  note         = {Machine review of arXiv:2412.08816}
}
read the original abstract

Neuromorphic applications emulate the processing performed by the brain by using spikes as inputs instead of time-varying analog stimuli. Therefore, these time-varying stimuli have to be encoded into spikes, which can induce important information loss. To alleviate this loss, some studies use population coding strategies to encode more information using a population of neurons rather than just one neuron. However, configuring the encoding parameters of such a population is an open research question. This work proposes an approach based on maximizing the mutual information between the signal and the spikes in the population of neurons. The proposed algorithm is inspired by the information-theoretic framework of Partial Information Decomposition. Two applications are presented: blood pressure pulse wave classification, and neural action potential waveform classification. In both tasks, the data is encoded into spikes and the encoding parameters of the neuron populations are tuned to maximize the encoded information using the proposed algorithm. The spikes are then classified and the performance is measured using classification accuracy as a metric. Two key results are reported. Firstly, adding neurons to the population leads to an increase in both mutual information and classification accuracy beyond what could be accounted for by each neuron separately, showing the usefulness of population coding strategies. Secondly, the classification accuracy obtained with the tuned parameters is near-optimal and it closely follows the mutual information as more neurons are added to the population. Furthermore, the proposed approach significantly outperforms random parameter selection, showing the usefulness of the proposed approach. These results are reproduced in both applications.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

42 extracted references · 32 canonical work pages

  1. [1]

    Predicting vascular age using PPG signals and machine learning algorithms: A non-invasive approach for early cardiovascular risk detection

    K. P. Abrisham et al. "Predicting vascular age using PPG signals and machine learning algorithms: A non-invasive approach for early cardiovascular risk detection". In: 2023 11th RSI International Conference on Robotics and Mechatronics (ICRoM). IEEE, Dec. 2023, pp. 72-77. DOI: 10 .1109/icrom60803. 2023 .10412455

    work page 2023

  2. [2]

    Feature representations for neuromorphic audio spike streams

    J. Anumula et al. "Feature representations for neuromorphic audio spike streams". In: Frontiers in Neuroscience 12 (Feb. 2018). DOI: 10. 3389/fnins. 2018. 00023. URL: https://doi.org/10.3389/fnins.2018.00023

    arXiv 2018

  3. [3]

    Neural correlations, population coding and computation

    B. B. Averbeck, P. E. Latham, and A. Pouget. "Neural correlations, population coding and computation". In: Nature Reviews Neuroscience 7.5 (May 2006), pp. 358-366. DOI: 10.1038/nrn1888. URL: https://doi.org/10.1038/nrn1888. REFERENCES 18

    doi:10.1038/nrn1888 2006

  4. [4]

    Tactile intensity and population codes

    S. J. Bensmaia. "Tactile intensity and population codes". In: Behavioural Brain Research 190.2 (July 2008), pp. 165-173. DOI: 10. 1016/ j . bbr. 2008. 02. 044. URL: https://doi.org/10.1016/j.bbr.2008.02.044

    work page doi:10.1016/j.bbr.2008.02.044 2008

  5. [5]

    LIBSVM: A library for support vector machines

    C.-C. Chang and C.-J. Lin. "LIBSVM: A library for support vector machines". In: ACM Transactions on Intelligent Systems and Technology 2.3 (Apr. 2011), pp. 1-27. DOI: 10 .1145/1961189 .1961199. URL: https: //doi. org/10 .1145/ 1961189 .1961199

    work page 2011

  6. [6]

    P. H. Charlton et al. Pulse Wave Database (PWDB): A database of arterial pulse waves representative of healthy adults. en. 2019. DOI: 10. 5281/ZEN0D0. 3275625. URL: https://zenodo.org/record/3275625

    work page arXiv 2019

  7. [7]

    Modeling arterial pulse waves in healthy aging: a database for in silico evaluation of hemodynamics and pulse wave indexes

    P. H. Charlton et al. "Modeling arterial pulse waves in healthy aging: a database for in silico evaluation of hemodynamics and pulse wave indexes". In: American Journal of Physiology-Heart and Circulatory Physiology 317.5 (Nov. 2019), H1062- H1085. DOI: 10.1152/ajpheart.00218.2019. URL: https://doi.org/10.1152/ ajpheart.00218.2019

    work page arXiv 2019

  8. [8]

    T. M. Cover and J. A. Thomas. Elements of information theory. 2nd. Wiley, 2006. DOI: 10.1002/047174882x. URL: https://doi.org/10.1002/047174882x

    doi:10.1002/047174882x 2006

Show all 42 references
  1. [9]

    Efficiency metrics for auditory neuro­ morphic spike encoding techniques using information theory

    A. El Ferdaoussi, J. Rouat, and E. Plourde. "Efficiency metrics for auditory neuro­ morphic spike encoding techniques using information theory". In: Neuromorphic Computing and Engineering 3.2 (June 2023), p. 024007. DOI: 10. 1088/ 2634- 4386/acd952. URL: https://doi.org/10.108...

    2023 doi

  2. [10]

    Optimizing neuromorphic spike encoding of dynamic stimulus signals using information theory

    A. El Ferdaoussi, J. Rouat, and E. Plourde. "Optimizing neuromorphic spike encoding of dynamic stimulus signals using information theory". In: 2023 11th International IEEE/EMBS Conference on Neural Engineering (NER). IEEE, Apr

    2023

  3. [11]

    Neuronal dynamics

    W. Gerstner et al. "Neuronal dynamics". In: Cambridge University Press, July

  4. [12]

    Bits and pieces: understanding information decomposition from part-whole relationships and formal logic

    A. J. Gutknecht, M. Wibral, and A. Makkeh. "Bits and pieces: understanding information decomposition from part-whole relationships and formal logic". In: Proceedings of the Royal Society A: Mathematical, Physical and Engineering Sciences 477.2251 (July 2021). ISSN: 1471-2946. ...

    2021

  5. [13]

    Basic response properties of auditory nerve fibers: a review

    P. Heil and A. J. Peterson. "Basic response properties of auditory nerve fibers: a review". In: Cell and Tissue Research 361.1 (Apr. 2015), pp. 129-158. DOI: 10.1007/s00441-015-2177-9. URL: https://doi.org/10.1007/s00441-015- 2177-9

    2015 doi

  6. [14]

    Information-theoretic methods for studying population codes

    R. A. Ince et al. "Information-theoretic methods for studying population codes". In: Neural Networks 23.6 (Aug. 2010), pp. 713-727. DOI: 10. 1016/ j . neunet. 2010.05.008. URL: https://doi.org/10.1016/j.neunet.2010.05.008. REFERENCES 19

    2010 doi

  7. [15]

    Synergy, redundancy, and independence in population codes, revisited

    P. E. Latham and S. Nirenberg. "Synergy, redundancy, and independence in population codes, revisited". In: The Journal of Neuroscience 25.21 (May 2005), pp. 5195-5206. DOI: 10. 1523/ jneurosci. 5319-04. 2005. URL: https: / / doi. org/10.1523/jneurosci.5319-04.2005

    2005 doi

  8. [16]

    Recent progress in multi-electrode spike sorting methods

    B. Lefebvre, P. Yger, and 0. Marre. "Recent progress in multi-electrode spike sorting methods". In: Journal of Physiology-Paris 110.4 (Nov. 2016), pp. 327-

    2016

  9. [17]

    Asynchronous binaural spatial audition sensor with 2x64x4 channel output

    S.-C. Liu et al. "Asynchronous binaural spatial audition sensor with 2x64x4 channel output". In: IEEE Transactions on Biomedical Circuits and Systems 8.4 (Aug. 2014), pp. 453-464. DOI: 10. 1109/tbcas. 2013. 2281834. URL: https: //doi.org/10.1109/tbcas.2013.2281834

    2014

  10. [18]

    Exploration of rank order coding with spiking neural networks for speech recognition

    S. Loiselle et al. "Exploration of rank order coding with spiking neural networks for speech recognition". In: Proceedings of the 2005 IEEE International Joint Conference on Neural Networks. IEEE. DOI: 10 .1109/ijcnn. 2005. 1556220. URL: https://doi.org/10.1109/ijcnn.2005.1556220

    2005

  11. [19]

    Send-On-Delta concept: An event-based data reporting strategy

    M. Miskowicz. "Send-On-Delta concept: An event-based data reporting strategy". In: Sensors 6.1 (Jan. 2006), pp. 49-63. DOI: 10. 3390/ s6010049. URL: https: //doi.org/10.3390/s6010049

    2006 doi

  12. [20]

    Revealing the Dynamics of Neural Information Processing with Multivariate Information Decomposition

    E. Newman et al. "Revealing the Dynamics of Neural Information Processing with Multivariate Information Decomposition". In: Entropy 24.7 (July 2022), p. 930. ISSN: 1099-4300. DOI: 10. 3390/e24070930. URL: http: //dx. doi. org/10. 3390/ e24070930

    2022

  13. [21]

    Predictive information in a sensory population

    S. E. Palmer et al. "Predictive information in a sensory population". In: Proceedings of the National Academy of Sciences 112.22 (May 2015), pp. 6908-

    2015

  14. [22]

    Neural population coding for effective temporal classification

    Z. Pan et al. "Neural population coding for effective temporal classification". In: 2019 International Joint Conference on Neural Networks (IJCNN). IEEE, July

    2019

  15. [23]

    Correcting for the sampling bias problem in spike train information measures

    S. Panzeri et al. "Correcting for the sampling bias problem in spike train information measures". In: Journal of Neurophysiology 98.3 (Sept. 2007), pp. 1064-1072. DOI: 10. 1152/ jn. 00559. 2007. URL: https: / / doi. org/ 10. 1152/jn.00559.2007

    2007

  16. [24]

    The structures and functions of correlations in neural population codes

    S. Panzeri et al. "The structures and functions of correlations in neural population codes". In: Nature Reviews Neuroscience 23.9 (June 2022), pp. 551-567. DOI: 10.1038/s41583-022-00606-4. URL: https://doi.org/10.1038/s41583-022- 00606-4

    2022 doi

  17. [25]

    Scikit-learn: Machine learning in Python

    F. Pedregosa et al. "Scikit-learn: Machine learning in Python". In: Journal of Machine Leaming Research 12 (2011), pp. 2825-2830. REFERENCES 20

    2011

  18. [26]

    How many neurons can we see with current spike sorting algorithms?

    C. Pedreira et al. "How many neurons can we see with current spike sorting algorithms?" In: Journal of Neuroscience Methods 211.1 (Oct. 2012), pp. 58-65. DOI: 10. 1016/ j . jneumeth. 2012. 07. 010. URL: https: / / doi. org/ 10. 1016/ j . jneumeth.2012.07.010

    2012

  19. [27]

    Inference and computation with population codes

    A. Pouget, P. Dayan, and R. S. Zemel. "Inference and computation with population codes". In: Annual Review of Neuroscience 26.1 (Mar. 2003), pp. 381-

    2003

  20. [28]

    Past, present and future of spike sorting techniques

    H. G. Rey, C. Pedreira, and R. Quian Quiroga. "Past, present and future of spike sorting techniques". In: Brain Research Bulletin 119 (Oct. 2015), pp. 106-117. ISSN: 0361-9230. DOI: 10. 1016 / j . brainresbull. 2015. 04. 007. URL: http: //dx.doi.org/10.1016/j.brainresbull.2015.04.007

    2015 doi

  21. [29]

    Mutual information between discrete and continuous data sets

    B. C. Ross. "Mutual information between discrete and continuous data sets". In: PLoS ONE 9.2 (2014). DOI: 10. 1371/ journal. pone. 0087357. URL: https: //doi.org/10.1371/journal.pone.0087357

    2014 doi

  22. [30]

    Synergy, redundancy, and independence in population codes

    E. Schneidman, W. Bialek, and M. J. Berry. "Synergy, redundancy, and independence in population codes". In: The Journal of Neuroscience 23.37 (Dec. 2003), pp. 11539-11553. DOI: 10. 1523/ jneurosci. 23-37-11539. 2003. URL: https://doi.org/10.1523/jneurosci.23-37-11539.2003

    2003

  23. [31]

    BSA, a fast and accurate spike train encoding scheme

    B. Schrauwen and J. Van Campenhout. "BSA, a fast and accurate spike train encoding scheme". In: Proceedings of the 2003 IEEE International Joint Conference on Neural Networks. Vol. 4. Institute of Electrical and Electronics Engineers (IEEE), 2003, pp. 2825-2830. DOI: 10 .1109/...

    2003

  24. [32]

    Entropy and information in neural spike trains

    S. P. Strong et al. "Entropy and information in neural spike trains". In: Physical Review Letters 80.1 (Jan. 1998), pp. 197-200. DOI: 10. 1103/physrevlett. 80

    1998

  25. [33]

    The Intel neuromorphic DNS challenge

    J. Timcheck et al. "The Intel neuromorphic DNS challenge". In: Neuromorphic Computing and Engineering (July 2023). DOI: 10 .1088/2634-4386/ace737. URL: https://doi.org/10.1088/2634-4386/ace737

    2023 doi

  26. [34]

    The upward bias in measures of information derived from limited data samples

    A. Treves and S. Panzeri. "The upward bias in measures of information derived from limited data samples". In: Neural Computation 7.2 (Mar. 1995), pp. 399-407. DOI: 10 .1162/neco .1995. 7. 2. 399. URL: https: //doi. org/10 .1162/neco. 1995.7.2. 399

    1995

  27. [35]

    P. L. Williams and R. D. Beer. Nonnegative decomposition of multivariate information. arXiv:1004.2515 [cs.IT]. 2010. DOI: 10. 48550/ ARXIV. 1004. 2515. arXiv: arXiv:1004.2515 [cs.IT]. URL: https://arxiv.org/abs/1004.2515

    2010 doi arXiv

  28. [197]

    URL: https://doi.org/10.1103/physrevlett.80.197

    doi

  29. [335]

    ISSN: 0928-4257. DOI: 10. 1016/ j . jphysparis. 2017. 02. 005. URL: http: //dx.doi.org/10.1016/j.jphysparis.2017.02.005

    2017 doi

  30. [410]

    1146/annurev

    DOI: 10. 1146/annurev. neuro. 26. 041002 .131112. URL: https: //doi. org/10.1146/annurev.neuro.26.041002.131112

  31. [2014]

    Chap. 5. DOI: 10 .1017 /cbo9781107447615. URL: https: //doi. org/10. 1017/cbo9781107447615

  32. [2019]

    DOI: 10 .1109/ijcnn. 2019. 8851858. URL: https: //doi. org/10. 1109/ ijcnn.2019.8851858

    2019

  33. [2023]

    DOI: 10 .1109/ner52421. 2023. 10123854. URL: https: //doi. org/10. 1109/ner52421.2023.10123854

    2023

  34. [6913]

    URL: https: //doi

    DOI: 10 .1073/pnas .1506855112. URL: https: //doi. org/10 .1073/pnas. 1506855112

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.