Pith. sign in

REVIEW 5 major objections 5 minor 30 references

Mixtures of Neural Cellular Automata: A Stochastic Framework for Growth Modelling and Self-Organization

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

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

Pith's one-line read Mixing several learned local update rules, selected cell by cell through a probability network, lets neural cellular automata represent stochastic tissue growth, recover from image perturbations, and segment cells without labels.

desk verdict Genuine but incremental extension of NCA with a learned rule selector and noise; the synthetic tissue results look good, but the biological claims outrun the evidence. read the letter →

arxiv 2506.20486 v1 pith:DNKXNCZM submitted 2025-06-25 cs.AI

Salvatore Milite , Giulio Caravagna , Andrea Sottoriva This is my paper · ORCID
classification cs.AI
keywords MixtureofNeuralCellularAutomatastochasticself-organizationtissuegrowthmodellingmorphogenesisunsupervisedsegmentationGumbel-Softmaxinterpretableruleassignment
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 proposes the Mixture of Neural Cellular Automata (MNCA), a stochastic version of the neural cellular automaton in which each cell, at each time step, draws one of K learned local update rules from a probability distribution produced by a small neural network. The authors claim that this probabilistic rule selection, optionally combined with injected Gaussian noise, captures random cell behavior, recovers better from image perturbations than a single deterministic rule, and yields interpretable rule assignments that segment cell populations and image regions. The evidence comes from three experiments: reproducing a synthetic agent-based model of tissue growth and differentiation, testing recovery of learned emoji and CIFAR-10 images after perturbations, and unsupervised segmentation of synthetic microscopy images. Because the biological benchmark is a hand-specified simulator rather than real tissue data, the paper's contribution is best read as a mechanism plus evidence on synthetic tasks, with the transfer to actual biological growth left as an assumption.

What carries the argument

The carrying mechanism is the mixture update rule. At each cell $i$ a rule-selector network $\pi$ maps the current state $s_i^t$ to a categorical distribution over $K$ local transition networks $\phi_k$. The active rule index $z$ is sampled from that distribution through the Gumbel-Softmax relaxation, and the next state is $s_i^{t+1}=s_i^t+\prod_{k=1}^K \phi_k(s_i^t,\{s_j^t\}_{j\in\mathcal{N}(i)},\mathbf{x}_k)^{z_k}$, where $\mathbf{x}_k\sim\mathcal{N}(0,1)$ is injected Gaussian noise in the noisy variant. The categorical selector imposes a soft clustering of cells into rule regimes, which is what produces segmentation and interpretability, while the sampling and noise turn the update into a stochastic transition kernel that can represent rare and cell-specific events.

What would settle it

Train an MNCA and a single-rule NCA of matched parameter count on time-lapse images of real growing tissue with stochastic lineage decisions, for instance organoid cultures tracked cell by cell; if the MNCA's KL divergence on cell-type composition and Wasserstein distances on tissue shape are not clearly better than the NCA's, the paper's central advantage claim falls. Alternatively, ablate the mixture by forcing all K rules to share identical weights: if perturbation recovery remains as good as a full mixture, then the mixture itself is not the mechanism.

Watch

Extended reading notes

Core claim

Read on its own terms, the paper's discovery is that replacing the single update rule of a neural cellular automaton with a stochastic mixture of K rules changes both what the model can represent and what can be read out of it. In the tissue-growth setup the MNCA lowers the KL divergence of cell-type proportions from 2.057 (NCA) to 0.018 (MNCA) and improves Wasserstein distances for tissue size and border structure; in morphogenesis the mixture recovers the target image after patch deletion, pixel removal, and Gaussian noise more accurately than the deterministic NCA and the Gaussian NCA baseline; and in microscopy images the learned rule probabilities correlate with a cell-shape parameter and can be re-weighted during inference to steer the population toward different morphologies. The authors interpret these outcomes as showing that MNCAs reproduce stochastic growth dynamics, provide interpretable rule segmentation, and are more stable under perturbations.

Load-bearing premise

The biological half of the central claim is tested only against a hand-written computer simulation of tissue growth, and the paper assumes that simulation is a faithful stand-in for real tissues; if that assumption fails, the claim that MNCAs recapitulate real biological growth patterns is unsupported.

Editorial extensions

If this is right

  • A trained MNCA can act as a black-box surrogate for a stochastic agent-based tissue model, reproducing cell-type proportions and spatial borders without hand-specified kinetic rates or ABC inference.
  • The perturbation results imply that mixture updates confer stability to learned morphogenesis without adversarial training, because no single rule determines the whole recovery trajectory.
  • Rule-probability maps are an emergent unsupervised segmentation of the input, and constraining rule probabilities at inference time steers the final phenotype.
  • The number of mixture rules is a capacity knob: fidelity improves with more rules but saturates, so practitioners can trade cost against divergence.

Reading between the lines

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

  • If the Lipschitz-averaging explanation is right, adding a categorical mixture of K copies of an already trained deterministic NCA should improve perturbation recovery even when the rules are not learned from biological data; this is a testable, biology-free prediction.
  • The paper's noise-tail analysis suggests the injected Gaussian channel acts as an interpretable latent variable for rare events; one could try to read out event types from the learned noise partition and use it as a controllable randomness knob in other self-organizing models.
  • Because the biological evaluation is synthetic, a natural next test is fitting MNCA to real time-lapse organoid or tumor imaging with lineage data; only that would settle whether the claimed recapitulation of real growth patterns holds outside the simulator.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

5 major / 5 minor

Summary. The paper proposes Mixture of Neural Cellular Automata (MNCA), an extension of NCA in which each cell's update is produced by one of K learned transition rules, selected probabilistically through a Gumbel-Softmax categorical sampler, optionally augmented with per-rule Gaussian noise. The framework is evaluated on three tasks: a synthetic tissue-growth simulator, emoji image morphogenesis under perturbations, and microscopic cell-image segmentation from synthetic high-content screening data. The authors claim that MNCAs are more robust to perturbations, better recapitulate biological growth patterns, and yield interpretable rule assignments, and they provide an appendix with an ABC comparison and a Lipschitz-based robustness argument. The core idea is clean and the experimental appendix is extensive, but several of the headline claims are broader than the reported evidence.

Significance. If the central claims hold after capacity controls and more careful evaluation, MNCAs are a useful extension of NCAs for stochastic self-organizing systems, with potential value in computational biology and generative modeling. The paper deserves credit for including an ABC baseline, a rule-number ablation, and CIFAR-10 generalization experiments in the appendices. The main tissue experiment shows a large improvement in cell-type distribution KL divergence relative to the single-rule NCA baseline. However, the robustness claim is contradicted in part of Table 3, the biological-dynamics claim is not supported by the marginal metrics used, and the mixture mechanism is not isolated from a parameter-count confound. These issues need to be resolved before the headline conclusions are acceptable.

major comments (5)
  1. [Section 4.3, Table 3] The blanket claim in Section 4.3 that 'Mixture-based NCA vastly outperforms the single-rule baseline across all perturbation scenarios' is contradicted by the first emoji block of Table 3: for Gaussian noise at 10% and 25%, the basic MNCA reports MSE 0.071±0.002 and 0.079±0.004, which are worse than the NCA's 0.049±0.007 and 0.052±0.010. Only MNCA with internal noise outperforms NCA in those conditions. Please report per-condition statistics and qualify the robustness claim accordingly.
  2. [Section 4.2, Appendix A and C, Section 5] The claim that MNCAs 'reproduce the stochastic dynamics observed in biological processes' is not supported by the reported metrics. The evaluation in Section 4.2 compares only marginal summaries—cell-type proportion KL, tissue-size Wasserstein, and border-size Wasserstein—of a hand-specified agent-based simulator (Algorithm 1). These marginals do not constrain transition probabilities, temporal correlations, or per-step stochastic behavior, even though the models are trained to reconstruct the state at each time step. Moreover, Section 5's reference to 'real spatial transcriptomic data' is unsupported: the microscopy experiment uses synthetic BBBC031 images, not real spatial transcriptomic data. Please add trajectory-level or transition-level metrics and real data, or temper the claims to synthetic surrogates.
  3. [Section 4.1, Table 2, Table 3] The MNCA baseline has K=5 update networks plus a rule selector, whereas the NCA baseline has a single update network, so part of the improvement in Tables 2 and 3 may reflect parameter count rather than the mixture mechanism. A matched-capacity control is needed—for example, a single NCA with proportionally wider layers, or an MNCA with K=1—before attributing the gains to probabilistic rule selection.
  4. [Appendix F] The theoretical robustness argument is not applied to the model actually evaluated. Equation (15) defines a convex-combination update F(s)=Σ π_k(s) f_k(s), but Eqs. (4b) and (5c) implement a hard categorical selection via the Gumbel-Softmax. The Lipschitz bound on the averaged update does not directly bound the stochastic selection process. Either revise the argument to address the sampled update or explicitly present it as a heuristic for why averaging may help.
  5. [Section 4.4, Figure 6C] The interpretable-segmentation claim rests on a visual 'clear correlation' between inferred rule assignments and the Cell Shape Parameter, but no quantitative correlation, statistical test, or agreement metric is reported. Since rule interpretability is one of the headline contributions, please quantify this relationship (e.g., correlation coefficient, adjusted Rand index against known phenotypes).
minor comments (5)
  1. [Section 3.1, Eq. (4)] The symbols z∼Cat(π(s_i,η)) and z_k are inconsistent: if z is a categorical value, z_k is undefined; if z is a one-hot vector, it should be defined as such. Please clarify the notation.
  2. [Section 4.1] The text lists 'Visium spatial transcriptomics' as one of the three experiments, but Section 4.4 uses BBBC031 microscopy images; please align the naming.
  3. [Table 2] The baseline is labeled 'GNCA' in Table 2 but 'GCA' in the text; please unify the notation.
  4. [Table 1] The table caption states 'k stands for 103'; this should be '10^3' or 'k=10^3'.
  5. [Appendix E, Table 4] The statement that ABC results are 'comparable' to the MNCA approach is not fully supported by Table 4: the ABC proportion-based model has Border-W 0.054 versus 0.184 for MNCA, so the comparison is mixed.

Circularity Check

1 steps flagged · score 2.0 of 10

No significant circularity: the MNCA derivation is self-contained; the only mild circular element is the K=5 rule count chosen to match the five known cell types in the synthetic tissue study, which partially pre-specifies the 'emergent' rule segmentation claim.

  1. fitted input called prediction [Section 4.2, rule-assignment paragraph near Figure 3]
    "When then looking at the rules assignment for the MNCAs, by plotting the rule assignment probability for a given state in time, we find that the model correctly gets that approximately each cell type has a different rule. This has also been our main rationale for setting the number of rules to 5."

    The number of mixture rules K is set to 5 because the synthetic tissue was generated with exactly five cell types (STEM, INT1, INT2, DIFF1, DIFF2). The paper then reports as an emergent result that approximately each cell type has a different rule. With five mixture components and five label classes, the coarse one-rule-per-cell-type alignment is a capacity choice rather than an independent discovery; the model cannot discover a different number of rule segments unless components are left unused or shared. The specific grouping (stem and DIFF2 sharing a rule, empty space owning a rule) is learned and not forced, so the circularity is partial and mild.

full rationale

The central MNCA formulation (Eq. 4-5) is a straightforward extension of NCAs with a categorical rule selector and Gaussian noise; no fitted parameter is renamed as a prediction. The robustness experiments in Section 4.3 and Appendix G train on clean targets and then test on unseen perturbations, so the improved recovery is an emergent external benchmark, not a self-fulfilling fit. The microscopy experiment in Section 4.4 correlates unsupervised rule assignments with the independent BBBC031 Cell Shape Parameter, providing external evidence for the segmentation claim. The only mild circular element is the choice K=5 in the synthetic tissue study, which matches the known five cell types and therefore partially pre-specifies the granularity of the claimed interpretable rule segmentation; however, the actual learned groupings deviate from the labels, so the observation retains unsupervised content. Appendix E explicitly acknowledges that the ABC baseline is evaluated on a quantity tied to its fitting statistic, an honest limitation of the comparison rather than a circularity in the MNCA derivation. Finally, the Conclusion's mention of 'real spatial transcriptomic data' is not supported by any experiment in the manuscript (only synthetic BBBC031 microscopy images are used); this is an evidence and reporting gap, not a circular reasoning step. Overall, the paper's core contributions are independently testable and not circular, with only minor pre-specification issues, so a low score is appropriate.

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

The central empirical claims rest on a hand-specified simulator as ground truth, standard reparameterization tricks, and unverified regularity and ergodicity assumptions in the robustness argument. The only notable free parameter is the number of rules, chosen post-hoc.

free parameters (3)
  • Number of mixture rules K = 5 for tissue and microscopy, 6 for emoji
    Chosen by hand based on the known number of cell types in the synthetic simulator, 5, and pilot performance. Appendix D shows KL decreases with K but with diminishing returns; the choice is post-hoc and affects interpretability results.
  • Gumbel-Softmax temperature = not reported
    The temperature schedule for the categorical reparameterization is not specified; it controls the entropy of rule selection and could affect training stability and interpretability.
  • Gaussian noise injection distribution = unit normal N(0,1)
    The scale of intrinsic noise is fixed to standard normal; the sensitivity of results to this scale is not explored.
assumptions (4)
  • standard math Gumbel-Softmax provides a differentiable approximation to sampling from the categorical rule distribution
    Used in Equation 4a to train the rule selector via backpropagation; standard result from Jang et al. (2016).
  • domain assumption The mixture update's Lipschitz constant is bounded by the weighted sum of component Lipschitz constants under mild regularity on the mixture weights
    Appendix F uses this to argue robustness through error averaging; the regularity conditions on the state-dependent weights are not stated or verified.
  • domain assumption The induced Markov chain over states is ergodic, or stochasticity helps escape attractors
    Appendix F argues randomness prevents lock-in; the paper admits the chain may not be ergodic in footnote 2, so the escape claim is not guaranteed.
  • domain assumption The synthetic agent-based tissue model in Appendix A is a valid proxy for real biological growth
    All quantitative biological-fidelity claims are measured against this simulator's outputs; no real biological data are used.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Mixtures of Neural Cellular Automata: A Stochastic Framework for Growth Modelling and Self-Organization." pith.science (2026). https://pith.science/paper/DNKXNCZM

@misc{pith2026250620486,
  author       = {Pith},
  title        = {Pith review of: Mixtures of Neural Cellular Automata: A Stochastic Framework for Growth Modelling and Self-Organization},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/DNKXNCZM}},
  note         = {Machine review of arXiv:2506.20486}
}
read the original abstract

Neural Cellular Automata (NCAs) are a promising new approach to model self-organizing processes, with potential applications in life science. However, their deterministic nature limits their ability to capture the stochasticity of real-world biological and physical systems. We propose the Mixture of Neural Cellular Automata (MNCA), a novel framework incorporating the idea of mixture models into the NCA paradigm. By combining probabilistic rule assignments with intrinsic noise, MNCAs can model diverse local behaviors and reproduce the stochastic dynamics observed in biological processes. We evaluate the effectiveness of MNCAs in three key domains: (1) synthetic simulations of tissue growth and differentiation, (2) image morphogenesis robustness, and (3) microscopy image segmentation. Results show that MNCAs achieve superior robustness to perturbations, better recapitulate real biological growth patterns, and provide interpretable rule segmentation. These findings position MNCAs as a promising tool for modeling stochastic dynamical systems and studying self-growth processes.

Figures

Figures reproduced from arXiv: 2506.20486 by the authors.

Figure 1
Figure 1. Architecture of the Mixture of Neural Cellular Automata (MNCA) model with noise injection. The model integrates signals from a central cell and its neighbours using spatial filters. The central cell’s features are processed by a Multi-Layer Perceptron (MLP) that implements a Rule Selector. The Rule Selector then determines probabilities for probabilistic Neural Cellular Automata (NCA) selection. The selected NCA is … view at source ↗
Figure 2
Figure 2. Visualization of MNCA’s stochastic framework applied to synthetic tissue growth. The figure illustrates the development of cellular patterns over time, starting from an initial configuration of stem cells and evolving into differentiated tissue structures. In panel A there is a brief description of the tissue model used for the simulation. In panel B we show an example of 8 realizations of the process. In the first … view at source ↗
Figure 3
Figure 3. Visualization of rule assignments in MNCA simulations On the first row, we split the tissue by cell type. On the second row, we have the rules assignment probability for each rule. Different rules are approximately assigned to distinct cell types. In contrast to the popular ABC-based ABM framework described in Section 2.1, MNCAs achieve comparable simulation quality with significantly less supervision (see Appendix … view at source ↗
Figures from the paper (13 more)
Figure 4
Figure 4. Figure 4: MNCA in image morphogenesis. The top half of the plot shows an example of the 3 perturbation types we studied in the paper. The columns are respectively, the original image, the perturbed image, the image after 50 steps of recovery, and the final recovered image after …
Figure 5
Figure 5. Figure 5: Visualization of rule assignments in MNCA on emojis. Different rules are assigned to distinct parts of the image. In particular, the body of the image, the empty space, and the borders tend to have specific rules. We used image set BBBC031v1 Piccinini et al. (2017), av…
Figure 6
Figure 6. Figure 6: MNCA fitting to cell microscopy images. A: Morphogenesis process initialized from fixed seeds, with one seed placed at each cell centroid in the input image. B: Probabilistic rule assignments for each cell, as inferred by the model. C: Correlation analysis between rule…
Figure 7
Figure 7. Figure 7: Steering the evolution of the system by changing the rule probabilities. A: Morphogenesis after reducing 100 times the probability of rule 0. The new image shows smaller and rounder cells. B: Morphogenesis after doubling the probability of the rule 5. Most of the irreg…
Figure 8
Figure 8. Figure 8: Comparative visualization of the vanilla MNCA versus the internal noise version. The top row is are trajectory from the training dataset. The middle row is the vanilla MNCA, and the last row is the MNCA with internal noise xt. Notably, the last row shows a higher perce…
Figure 9
Figure 9. Figure 9: Visualization of internal Gaussian noise distributions driving cell-state predictions in MNCA. The left panel shows the initial cell-state configuration, highlighting two specific spatial points. The center and right panels illustrate the noise-value distributions part…
Figure 10
Figure 10. Figure 10: KL divergence of the cell-type distribution as a function of the number of rules used in the Mixture NCA model. As the number of rules increases, the divergence between the simulated and target distributions decreases, indicating improved alignment with the expected c…
Figure 11
Figure 11. Figure 11: Simulated tissues by different agent-based models trained with ABC on the simulated dataset. Top row is 10 tissues from the training data. The second row are tissues generated by the ABC schema with type correlation as summary statistics. The third row are tissues gen…
Figure 12
Figure 12. Figure 12: Heatmap of each model’s product of linear layers singular.In our simple architecture, this product is an upper bound on the Lipschitz constant of the network. Lower values imply tighter Lipschitz bounds, suggesting greater stability under perturbations. While testing …
Figure 13
Figure 13. Figure 13: Final reconstructed image after perturbation for the deterministic NCA. We took as an example the perturbation with Gaussian injected noise in 25% of the pixels. This is the final image after 100 steps of recovery. A is the whale emoji (1F433) and B the avocado emoji …
Figure 14
Figure 14. Figure 14: Emoji morphogenesis for each NCA class. We show how all three 3 NCA classes converge to a visually good final image from the fixed seed at t=0. A, B, and C are respectively the deterministic NCA, the MNCA, and the MNCA with intrinsic noise. 27 [PITH_FULL_IMAGE:figure…
Figure 15
Figure 15. Figure 15: CIFAR-10 morphogenesis for each NCA class. We show how all three 3 NCA classes converge to a visually good final image from the fixed seed at t=0. A, B, and C are respectively the deterministic NCA, the MNC,A and the MNCA with intrinsic noise. 28 [PITH_FULL_IMAGE:fig…
Figure 16
Figure 16. Figure 16: Example of perturbation and recovery on CIFAR-10 dataset The top half of the plot shows an example of 2 perturbation types. The columns are respectively, the original image, the perturbed image, the image after 50 steps of recovery, and the final recovered image after…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

30 extracted references · 24 canonical work pages

  1. [1]

    Cellular automata models for vegetation dynamics

    Heiko Balzter, Paul W Braun, and Wolfgang K \"o hler. Cellular automata models for vegetation dynamics. Ecological modelling, 107 0 (2-3): 0 113--125, 1998

    work page 1998

  2. [2]

    Cell fate specification and differentiation in the adult mammalian intestine

    Joep Beumer and Hans Clevers. Cell fate specification and differentiation in the adult mammalian intestine. Nature reviews Molecular cell biology, 22 0 (1): 0 39--53, 2021

    work page 2021

  3. [3]

    Calibrating agent-based models to tumor images using representation learning

    Colin G Cess and Stacey D Finley. Calibrating agent-based models to tumor images using representation learning. PLoS Computational Biology, 19 0 (4): 0 e1011070, 2023

    work page 2023

  4. [4]

    Cellular automata approaches to biological modeling

    G Bard Ermentrout and Leah Edelstein-Keshet. Cellular automata approaches to biological modeling. Journal of theoretical Biology, 160 0 (1): 0 97--133, 1993

    work page 1993

  5. [5]

    Cellular automata as convolutional neural networks

    William Gilpin. Cellular automata as convolutional neural networks. Physical Review E, 100 0 (3): 0 032402, 2019

    work page 2019

  6. [6]

    Learning graph cellular automata

    Daniele Grattarola, Lorenzo Livi, and Cesare Alippi. Learning graph cellular automata. Advances in Neural Information Processing Systems, 34: 0 20983--20994, 2021

    work page 2021

  7. [7]

    Neural cellular automata manifold

    Alejandro Hernandez, Armand Vilalta, and Francesc Moreno-Noguer. Neural cellular automata manifold. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp.\ 10020--10028, 2021

    work page 2021

  8. [8]

    Categorical reparameterization with gumbel-softmax

    Eric Jang, Shixiang Gu, and Ben Poole. Categorical reparameterization with gumbel-softmax. arXiv preprint arXiv:1611.01144, 2016

    arXiv 2016

Show all 30 references
  1. [9]

    Parameter-efficient diffusion with neural cellular automata

    John Kalkhof, Arlene K \"u hn, Yannik Frisch, and Anirban Mukhopadhyay. Parameter-efficient diffusion with neural cellular automata. npj Unconventional Computing, 2 0 (1): 0 1--9, 2025

    2025

  2. [10]

    Adam: A method for stochastic optimization

    Diederik P Kingma. Adam: A method for stochastic optimization. arXiv preprint arXiv:1412.6980, 2014

    2014 arXiv

  3. [11]

    Computational framework for simulating fluorescence microscope images with cell populations

    Antti Lehmussola, Pekka Ruusuvuori, Jyrki Selinummi, Heikki Huttunen, and Olli Yli-Harja. Computational framework for simulating fluorescence microscope images with cell populations. IEEE transactions on medical imaging, 26 0 (7): 0 1010--1016, 2007

    2007

  4. [12]

    State-dependent evolutionary models reveal modes of solid tumour growth

    Maya A Lewinsohn, Trevor Bedford, Nicola F M \"u ller, and Alison F Feder. State-dependent evolutionary models reveal modes of solid tumour growth. Nature Ecology & Evolution, 7 0 (4): 0 581--596, 2023

    2023

  5. [13]

    Annotated high-throughput microscopy image sets for validation

    Vebjorn Ljosa, Katherine L Sokolnicki, and Anne E Carpenter. Annotated high-throughput microscopy image sets for validation. Nature methods, 9 0 (7): 0 637, 2012

    2012

  6. [14]

    Latent neural cellular automata for resource-efficient image restoration

    Andrea Menta, Alberto Archetti, and Matteo Matteucci. Latent neural cellular automata for resource-efficient image restoration. arXiv preprint arXiv:2403.15525, 2024

    2024 arXiv

  7. [15]

    Spectral normalization for generative adversarial networks

    Takeru Miyato, Toshiki Kataoka, Masanori Koyama, and Yuichi Yoshida. Spectral normalization for generative adversarial networks. arXiv preprint arXiv:1802.05957, 2018

    2018 arXiv

  8. [16]

    Growing neural cellular automata

    Alexander Mordvintsev, Ettore Randazzo, Eyvind Niklasson, and Michael Levin. Growing neural cellular automata. Distill, 5 0 (2): 0 e23, 2020

    2020

  9. [17]

    Spatial structure governs the mode of tumour evolution

    Robert Noble, Dominik Burri, C \'e cile Le Sueur, Jeanne Lemant, Yannick Viossat, Jakob Nikolas Kather, and Niko Beerenwinkel. Spatial structure governs the mode of tumour evolution. Nature ecology & evolution, 6 0 (2): 0 207--217, 2022

    2022

  10. [18]

    Variational neural cellular automata

    Rasmus Berg Palm, Miguel Gonz \'a lez-Duque, Shyam Sudhakaran, and Sebastian Risi. Variational neural cellular automata. arXiv preprint arXiv:2201.12360, 2022

    2022 arXiv

  11. [19]

    Advanced cell classifier: user-friendly machine-learning-based software for discovering phenotypes in high-content imaging data

    Filippo Piccinini, Tamas Balassa, Abel Szkalisity, Csaba Molnar, Lassi Paavolainen, Kaisa Kujala, Krisztina Buzas, Marie Sarazova, Vilja Pietiainen, Ulrike Kutay, et al. Advanced cell classifier: user-friendly machine-learning-based software for discovering phenotypes in high-...

    2017

  12. [20]

    Self-classifying mnist digits

    Ettore Randazzo, Alexander Mordvintsev, Eyvind Niklasson, Michael Levin, and Sam Greydanus. Self-classifying mnist digits. Distill, 5 0 (8): 0 e00027--002, 2020

    2020

  13. [21]

    Learning spatio-temporal patterns with neural cellular automata

    Alex D Richardson, Tibor Antal, Richard A Blythe, and Linus J Schumacher. Learning spatio-temporal patterns with neural cellular automata. PLOS Computational Biology, 20 0 (4): 0 e1011589, 2024

    2024

  14. [22]

    Morphogen gradients: from generation to interpretation

    Katherine W Rogers and Alexander F Schier. Morphogen gradients: from generation to interpretation. Annual review of cell and developmental biology, 27 0 (1): 0 377--407, 2011

    2011

  15. [23]

    Spatial biology of cancer evolution

    Zaira Seferbekova, Artem Lomakin, Lucy R Yates, and Moritz Gerstung. Spatial biology of cancer evolution. Nature Reviews Genetics, 24 0 (5): 0 295--313, 2023

    2023

  16. [24]

    Cancer stem cell tumor model reveals invasive morphology and increased phenotypical heterogeneity

    Andrea Sottoriva, Joost JC Verhoeff, Tijana Borovski, Shannon K McWeeney, Lev Naumov, Jan Paul Medema, Peter MA Sloot, and Louis Vermeulen. Cancer stem cell tumor model reveals invasive morphology and increased phenotypical heterogeneity. Cancer research, 70 0 (1): 0 46--56, 2010

    2010

  17. [25]

    Quantification of spatial subclonal interactions enhancing the invasive phenotype of pediatric glioma

    Haider Tari, Ketty Kessler, Nick Trahearn, Benjamin Werner, Maria Vinci, Chris Jones, and Andrea Sottoriva. Quantification of spatial subclonal interactions enhancing the invasive phenotype of pediatric glioma. Cell Reports, 40 0 (9), 2022

    2022

  18. [26]

    Attention-based neural cellular automata

    Mattie Tesfaldet, Derek Nowrouzezahrai, and Chris Pal. Attention-based neural cellular automata. Advances in Neural Information Processing Systems, 35: 0 8174--8186, 2022

    2022

  19. [27]

    Theory of self-reproducing automata

    John Von Neumann, Arthur Walter Burks, et al. Theory of self-reproducing automata. 1966

    1966

  20. [28]

    Learning to generate 3d shapes with generative cellular automata

    Dongsu Zhang, Changwoon Choi, Jeonghwan Kim, and Young Min Kim. Learning to generate 3d shapes with generative cellular automata. arXiv preprint arXiv:2103.04130, 2021

    2021 arXiv

  21. [29]

    Outdoor scene extrapolation with hierarchical generative cellular automata

    Dongsu Zhang, Francis Williams, Zan Gojcic, Karsten Kreis, Sanja Fidler, Young Min Kim, and Amlan Kar. Outdoor scene extrapolation with hierarchical generative cellular automata. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp.\ 20145--...

    2024

  22. [30]

    write newline

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

Pith tools

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