Building Deep, Equivariant Capsule Networks

R. Raghunatha Sarma; Sairaam Venkatraman; S. Balasubramanian

REVIEW 3 major objections 4 minor 17 references

Building Deep, Equivariant Capsule Networks

T0 review · 3 major / 4 minor · reviewed 2026-08-14 · deepseek-v4-flash

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

Pith's one-line read This paper claims that replacing pairwise capsule predictions with per-type equivariant functions and degree-centrality routing makes a capsule network provably equivariant, preserving its part-whole decomposition graph under…

desk verdict Solid routing-layer equivariance proof, but the full-network claim is underproved and the graph theorem is nearly definitional; worth a serious review for the capsule niche. read the letter →

arxiv 1908.01300 v3 pith:5ZTGUGJI submitted 2019-08-04 cs.LG cs.CVstat.ML

Sairaam Venkatraman , S. Balasubramanian , R. Raghunatha Sarma This is my paper · ORCID

classification cs.LGcs.CVstat.ML

keywords capsulenetworksequivariancegroup-equivariantconvolutionsspace-of-variationdegreecentralityroutingcompositionalitytransformationrobustnessSOVNET

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

Most capsule networks learn a separate prediction network for every pair of capsule types, which is expensive and hard to scale, and they lack formal guarantees that their representations transform predictably. This paper claims both problems can be fixed at once: replace pairwise predictions with a trainable equivariant function per capsule-type—a 'space-of-variation'—implemented with group-equivariant convolutions, and route by degree centrality in a graph of predictions. The paper proves that this specific routing layer is equivariant (Theorem 2.1) and that an equivariant routing procedure in its general family preserves the capsule-decomposition graph up to isomorphism under input transformations (Theorem 2.2). If correct, the result gives capsule networks a provable compositional invariance and makes deeper capsule architectures practical, because the number of prediction networks grows with capsule types rather than with all type pairs. Experiments on transformed MNIST, FashionMNIST, CIFAR-10, KMNIST, and SVHN show SOVNET beating several capsule baselines on transformation robustness and matching CNNs in some settings.

What carries the argument

The central object is the space-of-variation (SOV): a trainable, vector-valued function $\Psi_j^{l+1}$ on a group $G$ that each capsule-type uses to map shallower capsules into predictions for a deeper capsule, replacing the pairwise prediction networks of standard capsule nets. The layer is assembled from four equivariant pieces: the group-equivariant correlation $f_i^l \star \Psi_j^{l+1,p}$, the degree-centrality routing weights computed from cosine similarities among predictions, the squash non-linearity, and the weighted sum that produces deeper capsules. The paper defines the capsule-decomposition graph $G(x)$ whose vertices are capsules (type, position, pose, activation) and whose directed, weighted edges are routing coefficients; this graph is the formal object whose isomorphism under transformations is the paper's compositional guarantee.

What would settle it

Run a trained SOVNET on an input $x$ and on a transformed input $L_gx$ with $g$ in the group used by the convolutions, extract the routing coefficients $c_{ij}(g)$ from both runs, and check whether each edge weight obeys $c_{ij}(h^{-1}\circ g)$; a mismatch beyond floating-point tolerance would falsify Theorem 2.2 for the full architecture. A second check is layer-level: compare $([L_gF^l]\otimes \Psi_j^{l+1})$ with $L_g([F^l \otimes \Psi_j^{l+1}])$ for random input capsules and random equivariant filters.

Watch

Extended reading notes

Core claim

The central claim is that a capsule network can be built so that its routing layer is equivariant to a group $G$ of transformations, and that this equivariance is enough to preserve the compositional parse it computes. Theorem 2.1 states that for the layer defined in Algorithm 2, with group-equivariant correlation as the prediction mechanism and degree-centrality as the routing rule, $([L_gF^l]\otimes \Psi_j^{l+1}) = L_g([F^l \otimes \Psi_j^{l+1}])$: transforming the input first and then routing is the same as routing first and then transforming. Theorem 2.2 states that any network whose routing belongs to the summation family of Algorithm 1 and is equivariant has isomorphic capsule-decomposition graphs for $x$ and $L_gx$, where the isomorphism sends a capsule at group position $g$ to the same type at $h^{-1}\circ g$. The paper reads this as a formal guarantee that detected part-whole relationships, with their routing strengths, are preserved under transformations.

Load-bearing premise

The full-network guarantee depends on the primary capsule layer, the residual blocks, and the output pooling also being equivariant, but the paper proves equivariance only for the routing step in Algorithm 2.

Editorial extensions

If this is right

A layer needs one prediction network per capsule-type rather than one per pair of types, so the parameter count scales with the number of types, not the product of type counts.
Routing has no iteration-count hyperparameter, unlike dynamic routing and EM routing, so there is one less schedule to tune when building deeper models.
Because the layer is equivariant, training on geometrically transformed data should teach the same capsule decomposition at transformed positions, which is what the improved train-time augmentation results show.
The capsule-decomposition graph provides a concrete certificate: any equivariant routing rule from the Algorithm 1 family automatically yields isomorphic graphs under the group, so future routing rules can be checked by checking equivariance.

Reading between the lines

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

The theorem's scope is layer-level: the full-network guarantee is only as strong as the equivariance of the primary-capsule construction, residual blocks, and output pooling, none of which receives a proof in the paper.
The p4 group used in experiments is equivariant only to rotations in multiples of 90°, so SOVNET's improved accuracy on 30° and 60° test rotations is an empirical extrapolation, not something the proof promises.
A direct empirical test of Theorem 2.2—extract routing coefficients for $x$ and $L_gx$ and check the graph isomorphism—is not reported; it would be a cheap and decisive validation.
The same SOV design could in principle be combined with learned or continuous group grids (for example, finer roto-translation grids), and the routing rule would remain equivariant; the open question is the computational cost of the grid.

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.

Desk Editor's Note

read the letter

Two things to know before you read it: the equivariance proof for the routing layer (Algorithm 2) is correct and checkable, but the paper's headline claim — that the full SOVNET model is equivariant — is not actually proven; and Theorem 2.2 is more a formal restatement than a deep result. Despite that, the paper gives the capsule-network subfield two concrete things: it replaces the expensive pairwise prediction networks with group-equivariant convolutions, and it introduces a degree-centrality routing rule that has no iteration hyperparameter. The stepwise proof for Algorithm 2 is sound; it relies on standard GCNN equivariance and shows that the degree-score, softmax, and weighted sum all commute with the group action. That part is solid.

The soft spots start with the scope of the claim. The proof covers only the routing layer. The primary capsule construction, the modified residual blocks, and the output pooling are described qualitatively, and no proof shows they preserve the same group action. The output stage actually takes a max over group positions, which is invariant rather than equivariant. So Theorem 2.2, as written, does not apply to the full architecture described in Section 3. This is a genuine gap, though not a fatal one: the authors could either formalize the whole architecture or narrow the theorem to the routing subnetwork.

Theorem 2.2 is close to a tautology. The capsule-decomposition graph is built from the routing weights, so if routing is equivariant, the graph is equivariant by construction. It's a clean formalization of the intuition, but it doesn't add much beyond the definition.

Empirically, the paper is weaker. No code, no error bars, no ablations. The accuracy tables show SOVNET beating several capsule baselines on affine-transformed MNIST, FashionMNIST, and CIFAR-10, but there are odd points — e.g., on untransformed CIFAR-10, SOVNET drops from 88.3% to 47.6% under a 30° rotation, which the authors explain by noting that p4 only covers multiples of 90°. That's coherent, but it undercuts the broad "robustness" framing. Released code and standard deviations would make the empirical claims much more believable.

Who should read it: people actually building capsule networks or working on equivariant representation learning. For them, the degree-centrality routing and the routing-layer equivariance proof are worth engaging with. General readers can skim. It deserves a serious referee — the theoretical part is checkable and the direction is promising. I'd send it to review and ask for code, error bars, and a clear statement of which parts of the architecture the equivariance theorem actually covers.

Referee Report

3 major / 4 minor

Summary. The paper proposes SOVNET, a capsule-network framework in which each capsule type is associated with a trainable equivariant prediction function, instantiated with group-equivariant convolutions, and in which routing is performed by a degree-centrality-based aggregation. The paper proves that the routing layer (Algorithm 2) is equivariant (Theorem 2.1) and that any equivariant routing procedure in the family of Algorithm 1 preserves the capsule-decomposition graph up to isomorphism (Theorem 2.2). Experiments on affine-transformed MNIST, FashionMNIST, and CIFAR-10, plus KMNIST and SVHN, compare SOVNET against several capsule baselines and a few convolutional baselines.

Significance. If the full-model equivariance claim were established, this would be a useful contribution: the architecture is more parameter-efficient than pairwise prediction capsule networks, the stepwise equivariance proof for Algorithm 2 is structurally sound and relies only on standard GCNN results, and the graph-based formalization of compositionality is a worthwhile framing. The experimental study is broad and shows consistent gains in transformation robustness over the tested capsule baselines. The significance is tempered, however, by the gap between what is proved (the routing layer) and what is claimed in the abstract (that the particular model instance is equivariant), and by the near-tautological character of Theorem 2.2.

major comments (3)

[Section 3 / Theorem 2.1 scope] The central claim that the full SOVNET model is equivariant is not established. The proof of Theorem 2.1 in Appendix A.1 covers only the operations inside Algorithm 2: group-correlation predictions, degree-centrality scoring, weighted summation, and squash. Section 3 describes the primary capsule layer, the hidden SOVNET blocks, and the output capsule layer only qualitatively ('We only present an outline'), and no equivariance proof is given for these components. In particular, the output layer takes a maximum over group positions of the two-norm, which is an invariant operation; subgroup pooling in Cohen and Welling is invariant, not equivariant. If the output capsules are formed after this pooling, they do not satisfy the transformation law [L_g f](x)=f(g^{-1}x) that Theorem 2.2 requires. The authors should either define every component and prove its equivariance, or explicitly restrict the equivariance and graph-preservation claims to the internal layers and treat the final pooling as an invariant readout. This is load-bearing because the abstract and conclusion claim that 'this particular instance' is equivariant and hence preserves the compositional representation of an input.
[Appendix A.2 / Theorem 2.2] Theorem 2.2 is close to a tautology. After assuming that the routing procedure is equivariant, the proof's edge-preservation step simply restates this assumption in graph language ('due to the assumed equivariance of the model, ... is routed to ... with routing coefficient ...'), so the theorem adds no content beyond the definition of the graph. Furthermore, the graph definition assigns the same routing coefficient c^{l+1}_{ij}(g2) to every edge entering (i,g2) from any shallower position, so the graph does not represent position-specific part-whole relationships; the isomorphism then holds essentially by construction. The authors should either enrich the graph definition to include position-dependent edge weights or temper the claim that equivariance 'guarantees the preservation of detected compositionalities.' The proof also contains a typo: the transformed edge's first vertex is written as \tilde{f}^{l+1}_i(h^{-1}\circ g_1) but should be \tilde{f}^{l}_i(h^{-1}\circ g_1).
[Algorithm 2 / Appendix A.1] The proof of equivariance of DegreeScore requires division by the product of norms of prediction vectors. When a prediction vector is zero, which is a case the paper itself anticipates for unrelated capsule types in Section 1 (such capsules 'will project themselves to a vector of low activation'), the cosine similarity in the DegreeScore procedure is undefined. The appendix restricts to 'non-zero' maps, but Algorithm 2 as stated does not specify how to handle this case. The authors should specify a well-defined procedure for zero prediction vectors (for example, adding an epsilon to the denominator or defining the similarity to be zero when either norm vanishes) and state that this procedure remains equivariant for all inputs on which the network is defined.

minor comments (4)

[Algorithm 1] The summation index in the step defining f^{l+1}_j(g) starts at i=1 rather than i=0, which is inconsistent with the rest of the algorithm and with Algorithm 2.
[Table 4] Several entries in the CIFAR-10 table lack the percent sign (for example, '73.45' in the Capsnet row), which should be made uniform for readability.
[Section 3, output capsule layer] The sentence 'This is an equivariant operation, as it corresponds to the subgroup-pooling of (Cohen and Welling, 2016)' is inaccurate: subgroup pooling yields invariance, not equivariance, and this wording obscures the model-level issue described in the major comments.
[Section 3, architecture details] The 'modified residual block' and the primary capsule construction are described only in words, with no formal definition or hyperparameter details; since the code is 'to be released pending publication,' the experimental section as written is not fully reproducible.

Circularity Check

1 steps flagged · score 3.0 of 10

The graph-isomorphism theorem restates routing equivariance by construction; the layer-level equivariance proof itself is not circular.

self definitional [Section 2.2, definition of capsule-decomposition graph and Theorem 2.2; proof in Appendix A.2.]
"We construct an edge between capsules that are connected by routing, with the direction from the shallower capsule to the deeper capsule. Each of these edges are weighted by the corresponding routing coefficient. ... Further, assume that this routing procedure is equivariant with respect to the group G. Then, given an input x and ∀g ∈ G, G(x) and G([Lgx]) are isomorphic."

The edge set of G(x) is defined by the routing coefficients themselves: E(x) = {(tilde f_i^l(g1), tilde f_i^{l+1}(g2), c_ij^{l+1}(g2)) : g1 in Pool(g2)}. The theorem's hypothesis is precisely that the routing procedure is equivariant, meaning these coefficients satisfy c_ij^{l+1}(g) -> c_ij^{l+1}(h^{-1}g) when the input is transformed by L_h. The proof's isomorphism sends the edge with coefficient c_ij(g2) to the edge with coefficient c_ij(h^{-1}g2), which is exactly the assumed equivariance rewritten as graph isomorphism. Thus the conclusion is built into the definitions of the graph and of equivariant routing; it does not derive a new property. The genuinely independent part of the paper is Theorem A.1, which verifies each operation in Algorithm 2 is equivariant.

full rationale

No fitted parameter is presented as a prediction, and no self-citation is load-bearing; the references to Cohen and Welling are external, checkable results about group-equivariant convolutions. Theorem A.1 is a direct step-by-step equivariance proof for Algorithm 2 and is not circular. The circularity concern is limited to Theorem 2.2: the capsule-decomposition graph is defined out of the routing coefficients whose equivariance is assumed, so graph isomorphism is a restatement of the assumption rather than an independent consequence. The paper itself calls this 'quite unsurprising.' Separately, the model-level claim that the result applies to the full SOVNET architecture in Section 3 is not fully proved, since only Algorithm 2 is shown equivariant and the primary-capsule residual blocks, SELU, and group-position max-pooling are not given equivariance proofs; this is an omitted-proof gap, not a circularity. Overall, the derivation has one definitional/renaming step, while the layer-level equivariance content remains independent, giving a score of 3.

Assumptions & free parameters 2 free parameters · 5 assumptions · 1 invented entities

The equivariance proof rests on standard GCNN results and pointwise nonlinearities, not on fitted constants. The empirical claims rest on many hand-chosen architecture and training choices that are not fully specified, and the SOV notion is interpretive rather than independently testable.

free parameters (2)

Architecture hyperparameters (depth, capsule types and dimensions, group p4 vs p4m) = not fully reported
Hand-chosen per dataset in Section 3; these influence all reported accuracies and are not specified to a level that permits exact replication. They do not affect the abstract equivariance proof.
Margin loss schedule and reconstruction loss weight = 0.9/0.1/0.5 then 0.95/0.05/0.8; reconstruction weight 0.0005
Adopted from prior capsule papers and hand-set; they affect empirical results but are not used in the theoretical claim.

assumptions (5)

standard math Group correlation equivariance: [L_h f] ⋆ ψ = L_h(f ⋆ ψ), from Cohen and Welling (2016).
Used in Appendix Theorem A.1 to conclude that predictions S_{ij} are equivariant.
domain assumption Input images are modeled as functions on a group G, with the translation group a subgroup of G.
Stated in Section 2 before the equivariance definition; this is what allows p4 and p4m convolutions to be used.
domain assumption All nonlinearities used in the network are pointwise in the group coordinate and therefore equivariant.
The appendix uses this for softmax, squash, and degree scores; the primary-capsule residual blocks using SELU are only sketched and not formally verified.
domain assumption The routing pool Pool(g2) transforms covariantly under the group action.
Theorem A.2's edge correspondence assumes local routing neighborhoods map to neighborhoods under L_h; pooling is never formally defined.
ad hoc to paper Shallower capsules unrelated to a particular capsule type project to low activation.
Section 2 uses this expectation to motivate the SOV design, but it is not used in the equivariance proof and is not separately validated.

invented entities (1)

space-of-variation (SOV)
purpose: Conceptual label for the output of a capsule-type's prediction network; the paper claims it encodes the manifold of legal pose variations for that capsule type.
No falsifiable handle is given outside the paper. The SOV is an interpretation of the learned projection function and is not needed for the equivariance proof.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Building Deep, Equivariant Capsule Networks." pith.science (2026). https://pith.science/paper/5ZTGUGJI

@misc{pith2026190801300,
  author       = {Pith},
  title        = {Pith review of: Building Deep, Equivariant Capsule Networks},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/5ZTGUGJI}},
  note         = {Machine review of arXiv:1908.01300}
}

read the original abstract

Capsule networks are constrained by the parameter-expensive nature of their layers, and the general lack of provable equivariance guarantees. We present a variation of capsule networks that aims to remedy this. We identify that learning all pair-wise part-whole relationships between capsules of successive layers is inefficient. Further, we also realise that the choice of prediction networks and the routing mechanism are both key to equivariance. Based on these, we propose an alternative framework for capsule networks that learns to projectively encode the manifold of pose-variations, termed the space-of-variation (SOV), for every capsule-type of each layer. This is done using a trainable, equivariant function defined over a grid of group-transformations. Thus, the prediction-phase of routing involves projection into the SOV of a deeper capsule using the corresponding function. As a specific instantiation of this idea, and also in order to reap the benefits of increased parameter-sharing, we use type-homogeneous group-equivariant convolutions of shallower capsules in this phase. We also introduce an equivariant routing mechanism based on degree-centrality. We show that this particular instance of our general model is equivariant, and hence preserves the compositional representation of an input under transformations. We conduct several experiments on standard object-classification datasets that showcase the increased transformation-robustness, as well as general performance, of our model to several capsule baselines.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

17 extracted references · 14 canonical work pages

[1]

Afshar, A

P . Afshar, A. Mohammadi, and K. N. Plataniotis. Brain tumor t ype classiﬁcation via capsule networks. In 2018 25th IEEE International Conference on Image Processing (ICIP) , pages 3129–3133. IEEE,

work page 2018
[5]

HitNet: a neural network with capsules embedded in a Hit-or-Miss layer, extended with hybrid data augmentation and ghost capsules

A. Deliège, A. Cioppa, and M. V an Droogenbroeck. Hitnet: a ne ural network with capsules embedded in a hit-or-miss layer, extended with hybrid data augmentation and ghost cap sules. arXiv preprint arXiv:1806.06519 ,

work page Pith review arXiv
[7]

Group Equivariant Capsule Networks

URL https://github.com/mrjel/group_equivariant_capsules_pytorch. original-date: 2018-10- 05T13:23:40Z. 10 A PREPRINT - S EPTEMBER 27, 2019 J. E. Lenssen, M. Fey, and P . Libuschewski. Group equivarian t capsule networks. arXiv preprint arXiv:1806.05086 ,

work page Pith review arXiv 2018
[8]

original-date: 2018-04- 30T23:44:26Z

URL https://github.com/ethanleet/CapsNet. original-date: 2018-04- 30T23:44:26Z. Y . Netzer, T. Wang, A. Coates, A. Bissacco, B. Wu, and A. Y . Ng.Reading digits in natural images with unsupervised feature learning

work page 2018
[10]

DeepCaps: Goin g Deeper with Capsule Networks

J. Rajasegaran. Ofﬁcial Implementation of "DeepCaps: Goin g Deeper with Capsule Networks" paper (CVPR 2019).: brjathu/deepcaps, Sept

work page 2019
[11]

DeepCaps: Going Deeper with Capsule Networks

URL https://github.com/brjathu/deepcaps. original-date: 2019- 03-15T07:42:55Z. J. Rajasegaran, V . Jayasundara, S. Jayasekara, H. Jayasekara, S. Seneviratne, and R. Rodrigo. Deepcaps: Going deeper with capsule networks. arXiv preprint arXiv:1904.09546 ,

work page Pith review arXiv 2019
[12]

Sparse Unsupervised Capsules Generalize Better

D. Rawlinson, A. Ahmed, and G. Kowadlo. Sparse unsupervised capsules generalize better. arXiv preprint arXiv:1804.06094,

work page Pith review arXiv
[13]

V . M. d. Rosario, E. Borin, and M. Breternitz Jr. The multi-la ne capsule network (mlcn). arXiv preprint arXiv:1902.08431,

work page Pith review arXiv 1902

Show all 17 references

[14]

Tissera, K

D. Tissera, K. Kahatapitiya, R. Wijesinghe, S. Fernando, an d R. Rodrigo. Context-aware multipath networks. arXiv preprint arXiv:1907.11519,

1907 arXiv
[15]

H. Xiao, K. Rasul, and R. V ollgraf. Fashion-mnist: a novel im age dataset for benchmarking machine learning algo- rithms. arXiv preprint arXiv:1708.07747 ,

arXiv
[16]

L. Y ang. A PyTorch Implementation of Matrix Capsules with EM Routing: yl-1993/Matrix-Capsules-EM-PyTorch, Sept

1993
[17]

original-date: 2018-04- 21T07:33:47Z

URL https://github.com/yl-1993/Matrix-Capsules-EM-PyTor ch. original-date: 2018-04- 21T07:33:47Z. A Appendix We present proofs for the theorems mentioned in the main body . Theorem A.1. The SOVNET layer deﬁned in Algorithm 2, and denoted by the ope rator ⊗ as given above, sati...

1993
[18]

Each Sl+1 ij (g).Sl+1 kj (g) = ∑ dl+1−1 p=0 (f l i ⋆ Ψ l+1,p j )(g)(f l k ⋆ Ψ l+1,p j )(g). From the equivariance of ⋆, ∑ dl+1−1 p=0 ([Lhf l i ] ⋆ Ψ l+1,p j )(g)([Lhf l k] ⋆ Ψ l+1,p j )(g) = ∑ dl+1−1 p=0 Lh(f l i ⋆ Ψ l+1,p j )(g)Lh(f l k ⋆ Ψ l+1,p j )(g) = ∑ dl+1−1 p=0 (f l i ...

2019
[2011]

S. S. R. Phaye, A. Sikka, A. Dhall, and D. Bathula. Dense and di verse capsule networks: Making the capsules learn better. arXiv preprint arXiv:1805.04001 ,

arXiv
[2016]

T. S. Cohen, M. Geiger, J. Köhler, and M. Welling. Spherical c nns. arXiv preprint arXiv:1801.10130 ,

arXiv
[2017]

URL http://arxiv.org/abs/1709.01889. T. Clanuwat, M. Bober-Irizar, A. Kitamoto, A. Lamb, K. Y amamoto, and D. Ha. Deep learning for classical japanese literature. arXiv preprint arXiv:1812.01718 ,

arXiv
[2019]

original-date: 2019-07-31T16:16:24Z

URL https://github.com/HopefulRational/DeepCaps-PyTorch. original-date: 2019-07-31T16:16:24Z. T. Jeong, Y . Lee, and H. Kim. Ladder capsule network. In International Conference on Machine Learning , pages 3071–3079,

2019

Pith tools

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

[1] [1]

Afshar, A

P . Afshar, A. Mohammadi, and K. N. Plataniotis. Brain tumor t ype classiﬁcation via capsule networks. In 2018 25th IEEE International Conference on Image Processing (ICIP) , pages 3129–3133. IEEE,

work page 2018

[2] [5]

HitNet: a neural network with capsules embedded in a Hit-or-Miss layer, extended with hybrid data augmentation and ghost capsules

A. Deliège, A. Cioppa, and M. V an Droogenbroeck. Hitnet: a ne ural network with capsules embedded in a hit-or-miss layer, extended with hybrid data augmentation and ghost cap sules. arXiv preprint arXiv:1806.06519 ,

work page Pith review arXiv

[3] [7]

Group Equivariant Capsule Networks

URL https://github.com/mrjel/group_equivariant_capsules_pytorch. original-date: 2018-10- 05T13:23:40Z. 10 A PREPRINT - S EPTEMBER 27, 2019 J. E. Lenssen, M. Fey, and P . Libuschewski. Group equivarian t capsule networks. arXiv preprint arXiv:1806.05086 ,

work page Pith review arXiv 2018

[4] [8]

original-date: 2018-04- 30T23:44:26Z

URL https://github.com/ethanleet/CapsNet. original-date: 2018-04- 30T23:44:26Z. Y . Netzer, T. Wang, A. Coates, A. Bissacco, B. Wu, and A. Y . Ng.Reading digits in natural images with unsupervised feature learning

work page 2018

[5] [10]

DeepCaps: Goin g Deeper with Capsule Networks

J. Rajasegaran. Ofﬁcial Implementation of "DeepCaps: Goin g Deeper with Capsule Networks" paper (CVPR 2019).: brjathu/deepcaps, Sept

work page 2019

[6] [11]

DeepCaps: Going Deeper with Capsule Networks

URL https://github.com/brjathu/deepcaps. original-date: 2019- 03-15T07:42:55Z. J. Rajasegaran, V . Jayasundara, S. Jayasekara, H. Jayasekara, S. Seneviratne, and R. Rodrigo. Deepcaps: Going deeper with capsule networks. arXiv preprint arXiv:1904.09546 ,

work page Pith review arXiv 2019

[7] [12]

Sparse Unsupervised Capsules Generalize Better

D. Rawlinson, A. Ahmed, and G. Kowadlo. Sparse unsupervised capsules generalize better. arXiv preprint arXiv:1804.06094,

work page Pith review arXiv

[8] [13]

V . M. d. Rosario, E. Borin, and M. Breternitz Jr. The multi-la ne capsule network (mlcn). arXiv preprint arXiv:1902.08431,

work page Pith review arXiv 1902