Pith. sign in

REVIEW 4 major objections 6 minor 39 references

Nonlinear Neural Dynamics and Classification Accuracy in Reservoir Computing

T0 review · 4 major / 6 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read Near-linear reservoirs can solve nonlinear classification tasks.

desk verdict Genuinely novel empirical observation about weak nonlinearity in reservoirs, but the 'robust regime' depends on a fixed bias term and the paper needs code, error bars, and a bias sweep before the claim is solid. read the letter →

arxiv 2411.10047 v1 pith:FONM7AXF submitted 2024-11-15 cs.NE q-bio.NC

Claus Metzner , Achim Schilling , Andreas Maier , Patrick Krauss This is my paper · ORCID
classification cs.NEq-bio.NC MSC 68T0737D4592B20
keywords reservoircomputingrecurrentneuralnetworksweaklynonlineardynamicsedgeofchaosclassificationaccuracyprincipalcomponentanalysispseudoinversereadout
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

Reservoir computing is often expected to need strongly nonlinear neurons and rich spontaneous dynamics, but this paper argues that very little of either is required. Using a random recurrent network of tanh neurons with weak coupling and weak input signals, it shows that tasks such as separating points inside a circle from points outside it can be solved with near-perfect accuracy once a linear readout is trained with the pseudoinverse. The nonlinearity at work is so small that it is invisible in the leading principal components of the reservoir state and appears only in higher-order components. At stronger coupling, accuracy peaks near the edges of chaos, but the weakly coupled regime is more stable and often more accurate.

What carries the argument

The mechanism that carries the argument is the bias-shifted activation $\tanh(u+c)$, with random per-neuron biases of standard deviation $w'=0.1$. The constant $c$ shifts the operating point so that the Taylor expansion around the typical input contains a quadratic term, which is the minimal ingredient needed to turn the circular class boundary of the circle task into a linearly separable one; unbiased tanh has no such term. The other carrying element is the pseudoinverse-trained linear readout, which can detect changes in the reservoir state that are orders of magnitude smaller than the state itself, precisely the changes seen in higher-order principal components.

What would settle it

Run the same circle-task experiment with all reservoir biases set to zero while keeping $w=0.1$, $b=0$, and tanh neurons: the paper reports accuracy falls to chance, around 0.5. Any setup that keeps near-perfect circle accuracy without any bias-induced even-order term would refute the claim.

Watch

Extended reading notes

Core claim

On its own terms, the paper's central discovery is that a reservoir computer with ten tanh neurons, recurrent coupling strength $w=0.1$, and balanced random connectivity performs near-perfect nonlinear classification on the circle and XOR tasks, even though the neurons operate far from saturation and the difference between tanh and purely linear activations is of order $10^{-3}$. The class-separating information is carried by small higher-order Taylor terms of the biased activation function and becomes visible only in PCA components 3 and 4 of the reservoir state; the linear readout, fitted in one shot with the pseudoinverse, can exploit those weak signals. With stronger coupling ($w=0.5$), the accuracy curve as a function of excitatory/inhibitory balance shows two peaks at the boundaries between oscillatory/chaotic and chaotic/fixpoint dynamics, supporting the edge-of-chaos hypothesis, while chaotic dynamics at balanced coupling degrades performance. A related result is that the circle task requires an even, quadratic term in the activation function: unbiased tanh, whose Taylor series has only odd powers, fails unless random biases supply that quadratic component.

Load-bearing premise

The claim that extremely weak nonlinearity is enough rests on the random biases supplying a quadratic Taylor term: with the biases removed, the unbiased tanh activation has no quadratic term and the circle-task accuracy drops to chance.

Editorial extensions

If this is right

  • Weakly coupled reservoirs ($w=0.1$) can classify circle, XOR, and patches tasks with accuracies around 0.93–0.97, so strong recurrent coupling and chaotic dynamics are not prerequisites for nonlinear computation.
  • Input-related computations can ride on top of oscillatory or fixed-point attractors with little loss of accuracy; it is specifically chaotic dynamics that degrades performance.
  • At medium and strong coupling, accuracy as a function of balance shows two peaks near the oscillatory/chaotic and chaotic/fixed-point boundaries, supporting the edge-of-chaos hypothesis in this setting.
  • The task-relevant nonlinear effect is confined to higher-order principal components of the reservoir state, so analyses that keep only the leading variance components would discard exactly the information the readout uses.
  • A linear reservoir on the same tasks performs at chance level for circle and XOR, confirming that the small nonlinearity, not the recurrent structure alone, is doing the work.

Reading between the lines

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

  • The 'robust weakly nonlinear operating regime' depends on a source of even-order nonlinearity: for tanh neurons that source is the hand-set bias strength $w'=0.1$. If biases are removed or set to zero, the circle-task claim fails; other even activation functions would supply the same term.
  • A testable extension is to measure the second derivative of the effective activation at its operating point in a given recurrent network; this quantity, not the overall firing rate or variance, should predict whether a near-linear reservoir can perform nonlinear classification.
  • If higher-order PCA components matter this much, standard variance-based dimensionality reductions of neural recordings such as EEG or local field potentials may be discarding the very signals that carry nonlinear classifications.
  • The edge-of-chaos peaks occur only in strongly coupled reservoirs; in the weak-coupling regime there is no spontaneous chaos to be at the edge of, so the edge-of-chaos hypothesis is parameter-dependent rather than universal.
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 / 6 minor

Summary. The paper studies a tanh-neuron reservoir computer on synthetic classification tasks, controlling the reservoir's recurrent coupling strength w and excitatory/inhibitory balance b. The authors characterize reservoir dynamics with three measures—fluctuation F, temporal correlation C, and nonlinearity α—and relate them to classification accuracy. Their main claims are: (i) extremely weak nonlinearity, including the nearly linear activation y = s·tanh(u/s) with large s, suffices for tasks such as the circle and XOR tasks; (ii) accuracy can remain high even when oscillatory or fixpoint spontaneous dynamics dominate, whereas chaos degrades performance; and (iii) for stronger coupling, accuracy peaks near the oscillatory/chaotic and chaotic/fixpoint transitions, consistent with the edge-of-chaos hypothesis. The paper emphasizes a 'robust weakly nonlinear operating regime' and a role for higher-order principal components in carrying task-relevant information.

Significance. If fully established, the weakly nonlinear regime would be a noteworthy and practically useful finding: it would show that near-linear, weakly coupled reservoirs can solve nonlinear classification problems, and that readout layers can exploit very small higher-order signal components. The paper also provides a nuanced, task-dependent view of edge-of-chaos versus weak-coupling performance, which is of independent interest. Strengths include controlled comparisons (linear versus tanh neurons, independent test sets, readout trained by pseudoinverse) and dynamical measures that are defined independently of task accuracy. The empirical support is weakened, however, by the absence of error bars, the failure to quantify the claimed phase-boundary locations, and incomplete reproducibility: code and data are not provided, and the central weak-nonlinearity result depends on a fixed reservoir-bias strength that is never swept.

major comments (4)
  1. [Methods §2.2 and §3.5] The headline 'robust weakly nonlinear operating regime' is not established for unbiased tanh reservoirs. The bias strength is fixed at w'=0.1 (Methods §2.2) and is never varied, yet the authors themselves state in Section 3.5 that circle-task accuracy drops to chance when the reservoir biases are zero, because tanh(u) has no quadratic Taylor term. Thus the quadratic nonlinearity exploited in the circle task is supplied by the hand-chosen bias term tanh(u + c), not by the tanh activation alone. The paper should include a bias-strength sweep (for example w' = 0, 0.01, 0.03, 0.1, 0.3, 1) and, ideally, a direct test of the quadratic mechanism (e.g., comparing against a truncated Taylor model or measuring the curvature contribution). Without such evidence, the word 'robust' overstates what has been shown.
  2. [§3.5 and Fig. 5(b)] The reported accuracy at very large linearity parameters is surprising and needs a quantitative check. With y = s·tanh(u/s) and a bias of order c ≈ 0.1, the quadratic coefficient in the Taylor expansion around u = c scales as approximately -2c/s^2, which at s = 100 is about 2×10^-5. Figure 5(b) nevertheless reports accuracy around 0.9 in this regime, with the RMS reservoir activation below 0.2. This could be a genuine small-signal effect, but the authors should verify that it is not an artifact of numerical precision, training-set size, the specific bias realization, or the absence of output noise, and they should report error bars or repeated runs for this plot.
  3. [§3.7 and Fig. 6] The edge-of-chaos claim is not quantitatively tied to the dynamical phase boundaries. The text locates the accuracy peaks at b ≈ ±0.75 for w = 0.5, but Fig. 6 does not overlay the F and C measures (or the separately computed transition points) onto the accuracy curves for the same reservoirs, so the coincidence with the oscillatory/chaotic and chaotic/fixpoint transitions is asserted rather than demonstrated. The authors should quantify the transition locations from the F/C data and test whether the peaks occur at values statistically consistent with those transitions, particularly since the strong-coupling peak effect is most visible in the XOR task rather than across all five tasks.
  4. [General (Figs. 1, 3, 6)] Throughout the paper, accuracy and dynamical measures are reported as means over reservoirs without confidence intervals or variance information. This matters for the comparison between weak and medium coupling (w = 0.1 versus w = 0.3), where differences appear modest, and for the claim that oscillatory and fixpoint regimes 'ride on top' without much loss of accuracy. The authors should provide standard deviations, confidence bands, or significance tests, especially for the N = 10 reservoir results and for the N = 100 result in Fig. 6(f).
minor comments (6)
  1. [Abstract] There is a typo: 'classificiation' should be 'classification'.
  2. [§2.8] In the list of symbols, 'simuluation' should be 'simulation'.
  3. [§3.3] In the first paragraph of Section 3.3, 'N = −1' should be 'α = −1'.
  4. [§3.5 and Fig. 5(b)] The sentence 'the accuracy is actually best for linearity parameters around sopt ≈ 0' appears inconsistent with the following sentence and with the activation function, since s = 0 would make the output vanish and the text then says accuracy drops for smaller s. This is likely a typo for sopt ≈ 1; it should be corrected and the optimum location stated explicitly.
  5. [Eq. (3)] The argmax in Eq. (3) is written as arg max over n, but the output index is k (z_k); the notation should be consistent throughout.
  6. [Data availability (§5.4)] The statement that data and analysis programs 'will be made available upon reasonable request' is weaker than reproducible-research best practice. Given the surprising small-signal accuracy result, the authors should consider archiving the code and data alongside the paper.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the readout is fitted on training episodes and evaluated on independent test episodes, the dynamical measures F, C, alpha are defined from activation statistics independently of accuracy, and the disclosed bias dependence of the weak-nonlinearity result is a parameter contingency rather than a circular reduction.

full rationale

The paper's central claims are not circular by construction. The readout weights are computed by the pseudoinverse from reservoir states on a training data set and then evaluated on an independent test data set (Methods 2.3 and 2.7), so classification accuracy is not a fitted quantity. The dynamical measures F, C, and alpha are defined purely from the statistics of neural activations (Methods 2.4-2.6), without any reference to task labels or readout accuracy; the phase boundaries in Figs. 1 and 6 are therefore not constructed from accuracy. The weak-nonlinearity result is supported by controlled comparisons of tanh versus linear reservoirs on identical inputs and initial conditions (Sec. 3.5-3.6), and the PCA analysis is a post-hoc visualization, not part of the readout construction. The one caveat the paper itself discloses -- that unbiased tanh neurons fail on the circle task because tanh(u) has no quadratic Taylor term, while biased tanh(u+c) does -- is an explicit mechanism statement, not a hidden circularity: the bias strength w' = 0.1 is a fixed architectural parameter chosen before training and is not fitted to test accuracy, and the Taylor-expansion argument can be checked independently. Self-citations [21,29] motivate the choice of w and b for controlling dynamics, but the paper recomputes the dynamical behavior directly in Fig. 1, so the cited prior work is not load-bearing for the reported predictions. Concerns about numerical precision, the absence of a bias-strength sweep, or the robustness of the s=100 result are scientific validity issues, not circularity.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

The model is a standard tanh reservoir; no new entities are postulated. The main hand-chosen inputs are bias strength and coupling strengths. The edge-of-chaos peak interpretation rests on an assumed but not directly shown alignment between accuracy peaks and phase boundaries.

free parameters (3)
  • reservoir bias strength w' = 0.1
    Set by hand; the circle task fails if biases are zero because tanh(u) lacks a quadratic Taylor term (Section 3.5).
  • recurrent coupling strengths w = 0.1, 0.3, 0.5, plus size-adapted values for N=100
    Three representative hand-chosen values spanning weak to strong coupling; the peak locations and accuracy depend on these choices.
  • density d = 1 (fully connected)
    Fixed to maximum throughout; the role of sparsity is not explored.
assumptions (4)
  • standard math The pseudoinverse readout gives the optimal linear mapping from reservoir states to class labels.
    Used to train the readout in Section 2.3; relies on SVD and the Moore-Penrose inverse.
  • domain assumption The dynamical regime can be classified by the measures F, C, and alpha as defined in Sections 2.4-2.6.
    The paper identifies oscillatory, chaotic, and fixed-point regimes using these aggregates without a formal order parameter or Lyapunov spectrum.
  • domain assumption The artificial tasks (line, circle, XOR, patches, spatio-temporal) are representative probes of reservoir computation.
    Results are claimed generally, but only these toy tasks are tested.
  • ad hoc to paper Accuracy peaks near b approximately plus or minus 0.75 coincide with the oscillatory/chaotic and chaotic/fixed-point transitions.
    Used to support the edge-of-chaos claim in Section 3.7 and Figure 6; no quantitative overlay of the F/C phase boundaries with accuracy curves is provided.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Nonlinear Neural Dynamics and Classification Accuracy in Reservoir Computing." pith.science (2026). https://pith.science/paper/FONM7AXF

@misc{pith2026241110047,
  author       = {Pith},
  title        = {Pith review of: Nonlinear Neural Dynamics and Classification Accuracy in Reservoir Computing},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/FONM7AXF}},
  note         = {Machine review of arXiv:2411.10047}
}
read the original abstract

Reservoir computing - information processing based on untrained recurrent neural networks with random connections - is expected to depend on the nonlinear properties of the neurons and the resulting oscillatory, chaotic, or fixpoint dynamics of the network. However, the required degree of nonlinearity and the range of suitable dynamical regimes for a given task are not fully understood. To clarify these questions, we study the accuracy of a reservoir computer in artificial classification tasks of varying complexity, while tuning the neuron's degree of nonlinearity and the reservoir's dynamical regime. We find that, even for activation functions with extremely reduced nonlinearity, weak recurrent interactions and small input signals, the reservoir is able to compute useful representations, detectable only in higher order principal components, that render complex classificiation tasks linearly separable for the readout layer. When increasing the recurrent coupling, the reservoir develops spontaneous dynamical behavior. Nevertheless, the input-related computations can 'ride on top' of oscillatory or fixpoint attractors without much loss of accuracy, whereas chaotic dynamics reduces task performance more drastically. By tuning the system through the full range of dynamical phases, we find that the accuracy peaks both at the oscillatory/chaotic and at the chaotic/fixpoint phase boundaries, thus supporting the 'edge of chaos' hypothesis. Our results, in particular the robust weakly nonlinear operating regime, may offer new perspectives both for technical and biological neural networks with random connectivity.

Figures

Figures reproduced from arXiv: 2411.10047 by the authors.

Figure 1
Figure 1. Reservoir computer and free reservoir dynamics. (a) In every time step t of an ongoing computation, an input matrix I of size N×M couples M input signals xm into the reservoir. The N neurons of the reservoir, which are recurrently connected by a matrix W of size N ×N, produce neural activations yn. An output matrix O of size K ×N reads the reservoir states yn and linearly extracts from them K output features zk. The… view at source ↗
Figure 2
Figure 2. Model classification tasks and states of the running RC. The ’purely spatial’ discrimination tasks (a-d) have only M =2 simultaneous input signals x0 ∈ [−1, +1] and x1 ∈ [−1, +1] per episode, and thus each input vector (x0, x1) can be represented as a point in a two-dimensional plane, with point colors representing the classes. By contrast, task (e) is a ’spatio-temporal’ pattern recognition task. (a): ’Line task’: … view at source ↗
Figure 3
Figure 3. Reservoir dynamics with input. The top panels (a-c) show the three quantitative statistical measures F (blue), C (orange) and α (green) of the reservoir dynamics as functions of the balance b, for three different recurrent couplings w, both without external input signals (solid lines) and with continuously fed-in spatio-temporal input signals (dashed lines). (a): For weak coupling w = 0.1, there is no significant st… view at source ↗
Figures from the paper (3 more)
Figure 4
Figure 4. Figure 4: Classification with readout, but no reservoir. The performance of the isolated readout layer is tested with four purely spatial, binary classification tasks. Top row: input data distributions p(x); Middle row: distribution of the readout layer’s linear output p(z), bef…
Figure 5
Figure 5. Figure 5: Effect of neural nonlinearity in the RC. (a): Modified neural activation function with a tunable linearity parameter s. For s > 1 (red curve), the range of arguments u where the neuron response is approximately linear is extended, compared to the regular tanh function …
Figure 6
Figure 6. Figure 6: Effect of reservoir dynamics on RC performance. Panels (a)-(e) show the accuracy of a RC (reservoir of N =10 neurons with tanh activation) in different tasks, as functions of the exci￾tatory/inhibitory balance b, for three different recurrent coupling strengths w=0.1 (…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

39 extracted references · 34 canonical work pages

  1. [1]

    Deep learning

    Yann LeCun, Yoshua Bengio, and Geoffrey Hinton. Deep learning. nature, 521(7553):436–444, 2015

    2015

  2. [2]

    Re- view of deep learning: Concepts, cnn architectures, challenges, applications, future directions

    Laith Alzubaidi, Jinglan Zhang, Amjad J Humaidi, Ayad Al-Dujaili, Ye Duan, Omran Al- Shamma, Jos´ e Santamar ´ ıa, Mohammed A Fadhel, Muthana Al-Amidie, and Laith Farhan. Re- view of deep learning: Concepts, cnn architectures, challenges, applications, future directions. Journal of big Data , 8(1):1–74, 2021

    work page 2021

  3. [3]

    Recent advances in natural language pro- cessing via large pre-trained language models: A survey

    Bonan Min, Hayley Ross, Elior Sulem, Amir Pouran Ben Veyseh, Thien Huu Nguyen, Oscar Sainz, Eneko Agirre, Ilana Heintz, and Dan Roth. Recent advances in natural language pro- cessing via large pre-trained language models: A survey. ACM Computing Surveys , 56(2):1–40, 2023

    work page 2023

  4. [4]

    Universality and individuality in neural dynamics across large populations of recurrent networks

    Niru Maheswaranathan, Alex H Williams, Matthew D Golub, Surya Ganguli, and David Sussillo. Universality and individuality in neural dynamics across large populations of recurrent networks. Advances in neural information processing systems , 2019:15629, 2019

    work page 2019

  5. [5]

    Recurrent neural networks are univer- sal approximators

    Anton Maximilian Sch¨ afer and Hans Georg Zimmermann. Recurrent neural networks are univer- sal approximators. In International Conference on Artificial Neural Networks , pages 632–640. Springer, 2006

    work page 2006

  6. [6]

    Universal approximation of flows of control systems by recurrent neural networks

    Miguel Aguiar, Amritam Das, and Karl H Johansson. Universal approximation of flows of control systems by recurrent neural networks. In 2023 62nd IEEE Conference on Decision and Control (CDC), pages 2320–2327. IEEE, 2023

    work page 2023

  7. [7]

    echo state

    Herbert Jaeger. The “echo state” approach to analysing and training recurrent neural networks- with an erratum note. Bonn, Germany: German National Research Center for Information Technology GMD Technical Report, 148(34):13, 2001

    work page 2001

  8. [8]

    Optimal sequence memory in driven random networks

    Jannis Schuecker, Sven Goedeke, and Moritz Helias. Optimal sequence memory in driven random networks. Physical Review X , 8(4):041029, 2018

    work page 2018

Show all 39 references
  1. [9]

    Connectivity, dynamics, and memory in reservoir computing with binary and analog neurons

    Lars B¨ using, Benjamin Schrauwen, and Robert Legenstein. Connectivity, dynamics, and memory in reservoir computing with binary and analog neurons. Neural computation, 22(5):1272–1311, 2010

    2010

  2. [10]

    Information process- ing capacity of dynamical systems

    Joni Dambre, David Verstraeten, Benjamin Schrauwen, and Serge Massar. Information process- ing capacity of dynamical systems. Scientific reports, 2(1):1–7, 2012

    2012

  3. [11]

    Randomly connected networks have short temporal memory

    Edward Wallace, Hamid Reza Maei, and Peter E Latham. Randomly connected networks have short temporal memory. Neural computation, 25(6):1408–1439, 2013

    2013

  4. [12]

    Fading memory echo state networks are universal

    Lukas Gonon and Juan-Pablo Ortega. Fading memory echo state networks are universal. Neural Networks, 138:10–13, 2021

    2021

  5. [13]

    Gradient-based learning drives robust representations in recurrent neural networks by balancing compression and expansion

    Matthew Farrell, Stefano Recanatesi, Timothy Moore, Guillaume Lajoie, and Eric Shea-Brown. Gradient-based learning drives robust representations in recurrent neural networks by balancing compression and expansion. Nature Machine Intelligence , 4(6):564–573, 2022

    2022

  6. [14]

    Stimulus-dependent suppression of chaos in recurrent neural networks

    Kanaka Rajan, LF Abbott, and Haim Sompolinsky. Stimulus-dependent suppression of chaos in recurrent neural networks. Physical Review E , 82(1):011903, 2010

    2010

  7. [15]

    Controlling recurrent neural networks by conceptors

    Herbert Jaeger. Controlling recurrent neural networks by conceptors. arXiv preprint arXiv:1403.3369, 2014. 20

    2014 arXiv

  8. [16]

    Understanding and controlling memory in recurrent neural networks

    Doron Haviv, Alexander Rivkind, and Omri Barak. Understanding and controlling memory in recurrent neural networks. In International Conference on Machine Learning, pages 2663–2671. PMLR, 2019

    2019

  9. [17]

    Suppressing chaos in neural networks by noise

    Lutz Molgedey, J Schuchhardt, and Heinz G Schuster. Suppressing chaos in neural networks by noise. Physical review letters , 69(26):3717, 1992

    1992

  10. [18]

    Noise-modulated neural networks as an application of stochastic resonance

    Shuhei Ikemoto, Fabio DallaLibera, and Koh Hosoda. Noise-modulated neural networks as an application of stochastic resonance. Neurocomputing, 277:29–37, 2018

    2018

  11. [19]

    Recurrence resonance” in three-neuron motifs

    Patrick Krauss, Karin Prebeck, Achim Schilling, and Claus Metzner. Recurrence resonance” in three-neuron motifs. Frontiers in computational neuroscience, 13, 2019

    2019

  12. [20]

    Control of noise- induced coherent oscillations in time-delayed neural motifs

    Florian B¨ onsel, Patrick Krauss, Claus Metzner, and Marius E Yamakou. Control of noise- induced coherent oscillations in time-delayed neural motifs. arXiv preprint arXiv:2106.11361 , 2021

    2021 arXiv

  13. [21]

    Dynamics and information import in recurrent neural net- works

    Claus Metzner and Patrick Krauss. Dynamics and information import in recurrent neural net- works. Frontiers in Computational Neuroscience, 16, 2022

    2022

  14. [22]

    Recurrent neural networks as versatile tools of neuroscience research

    Omri Barak. Recurrent neural networks as versatile tools of neuroscience research. Current opinion in neurobiology , 46:1–6, 2017

    2017

  15. [23]

    Highly nonrandom features of synaptic connectivity in local cortical circuits

    Sen Song, Per Jesper Sj¨ ostr¨ om, Markus Reigl, Sacha Nelson, and Dmitri B Chklovskii. Highly nonrandom features of synaptic connectivity in local cortical circuits. PLoS biology, 3(3):e68, 2005

    2005

  16. [24]

    Is cortical connectivity optimized for storing information? Nature neuroscience, 19(5):749–755, 2016

    Nicolas Brunel. Is cortical connectivity optimized for storing information? Nature neuroscience, 19(5):749–755, 2016

    2016

  17. [25]

    Exploring sparsity in recurrent neural networks

    Sharan Narang, Erich Elsen, Gregory Diamos, and Shubho Sengupta. Exploring sparsity in recurrent neural networks. arXiv preprint arXiv:1704.05119 , 2017

    2017 arXiv

  18. [26]

    Sparsity through evolutionary pruning prevents neuronal networks from overfitting

    Richard C Gerum, Andr´ e Erpenbeck, Patrick Krauss, and Achim Schilling. Sparsity through evolutionary pruning prevents neuronal networks from overfitting. Neural Networks , 128:305– 312, 2020

    2020

  19. [27]

    Effect of dilution in asym- metric recurrent neural networks

    Viola Folli, Giorgio Gosti, Marco Leonetti, and Giancarlo Ruocco. Effect of dilution in asym- metric recurrent neural networks. Neural Networks, 104:50–59, 2018

    2018

  20. [28]

    Analysis of structure and dynamics in three-neuron motifs

    Patrick Krauss, Alexandra Zankl, Achim Schilling, Holger Schulze, and Claus Metzner. Analysis of structure and dynamics in three-neuron motifs. Frontiers in Computational Neuroscience , 13:5, 2019

    2019

  21. [29]

    Weight statistics controls dynamics in recurrent neural networks.PloS one, 14(4):e0214541, 2019

    Patrick Krauss, Marc Schuster, Verena Dietrich, Achim Schilling, Holger Schulze, and Claus Met- zner. Weight statistics controls dynamics in recurrent neural networks.PloS one, 14(4):e0214541, 2019

    2019

  22. [30]

    Intrinsic noise improves speech recognition in a computational model of the auditory pathway

    Achim Schilling, Richard Gerum, Claus Metzner, Andreas Maier, and Patrick Krauss. Intrinsic noise improves speech recognition in a computational model of the auditory pathway. Frontiers in Neuroscience, 16:908330, 2022

    2022

  23. [31]

    Stochastic resonance controlled upregulation of internal noise after hearing loss as a putative cause of tinnitus-related neuronal hyperactivity

    Patrick Krauss, Konstantin Tziridis, Claus Metzner, Achim Schilling, Ulrich Hoppe, and Holger Schulze. Stochastic resonance controlled upregulation of internal noise after hearing loss as a putative cause of tinnitus-related neuronal hyperactivity. Frontiers in neuroscience, 1...

    2016

  24. [32]

    The stochastic res- onance model of auditory perception: A unified explanation of tinnitus development, zwicker tone illusion, and residual inhibition

    Achim Schilling, Konstantin Tziridis, Holger Schulze, and Patrick Krauss. The stochastic res- onance model of auditory perception: A unified explanation of tinnitus development, zwicker tone illusion, and residual inhibition. Progress in Brain Research, 262:139–157, 2021

    2021

  25. [33]

    Predictive coding and stochastic resonance as fundamental principles of auditory phantom perception

    Achim Schilling, William Sedley, Richard Gerum, Claus Metzner, Konstantin Tziridis, Andreas Maier, Holger Schulze, Fan-Gang Zeng, Karl J Friston, and Patrick Krauss. Predictive coding and stochastic resonance as fundamental principles of auditory phantom perception. Brain, 146...

    2023

  26. [34]

    Recurrence resonance– noise-enhanced dynamics in recurrent neural networks

    Claus Metzner, Achim Schilling, Andreas Maier, and Patrick Krauss. Recurrence resonance– noise-enhanced dynamics in recurrent neural networks. arXiv preprint arXiv:2408.05579 , 2024

    2024 arXiv

  27. [35]

    Real-time computing without stable states: A new framework for neural computation based on perturbations

    Wolfgang Maass, Thomas Natschl¨ ager, and Henry Markram. Real-time computing without stable states: A new framework for neural computation based on perturbations. Neural compu- tation, 14(11):2531–2560, 2002

    2002

  28. [36]

    What are extreme learning machines? filling the gap between frank rosen- blatt’s dream and john von neumann’s puzzle

    Guang-Bin Huang. What are extreme learning machines? filling the gap between frank rosen- blatt’s dream and john von neumann’s puzzle. Cognitive Computation, 7:263–278, 2015

    2015

  29. [37]

    Dy- namical coding of sensory information with competitive networks

    MI Rabinovich, R Huerta, A Volkovskii, HDI Abarbanel, M Stopfer, and Gilles Laurent. Dy- namical coding of sensory information with competitive networks. Journal of Physiology-Paris , 94(5-6):465–471, 2000

    2000

  30. [38]

    Computation at the edge of chaos: Phase transitions and emergent compu- tation

    Chris G Langton. Computation at the edge of chaos: Phase transitions and emergent compu- tation. Physica D: Nonlinear Phenomena , 42(1-3):12–37, 1990

    1990

  31. [39]

    Do reservoir computers work best at the edge of chaos? Chaos: An Interdisciplinary Journal of Nonlinear Science , 30(12), 2020

    Thomas L Carroll. Do reservoir computers work best at the edge of chaos? Chaos: An Interdisciplinary Journal of Nonlinear Science , 30(12), 2020. 22

    2020

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.