Pith. sign in

REVIEW 4 major objections 5 minor 87 references

Representation Learning for Equivariant Inference with Guarantees

T0 review · 4 major / 5 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read Symmetry-aware spectral contrastive learning yields equivariant conditional models with non-asymptotic error bounds and an effective sample size of $d_{\mathrm{iso}}N$.

desk verdict Novel equivariant contrastive architecture with a main theorem that overclaims; the proof requires extra conditions that are not stated, so the advertised statistical guarantees fail for heavy-tailed observables. read the letter →

arxiv 2505.19809 v3 pith:AI5WRDJ5 submitted 2025-05-26 cs.LG cs.AIcs.RO

Daniel Ordoñez-Apraez , Vladimir Kostić , Alek Fröhlich , Vivien Brandt , Karim Lounici , Massimiliano Pontil This is my paper
classification cs.LGcs.AIcs.RO MSC 68T0762G05
keywords equivariantrepresentationlearningconditionalexpectationoperatorspectralcontrastiveisotypicdecompositionnon-asymptoticstatisticalguaranteesuncertaintyquantificationrareeventsgroupsymmetry
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 aims to show that a known symmetry prior on the data-generating process is enough to turn spectral representation learning for conditional distributions into a provably sample-efficient method. It introduces an equivariant version of Neural Conditional Probability (eNCP): a G-equivariant, block-diagonal bilinear model that approximates the truncated singular value decomposition of the conditional expectation operator, with one block per isotypic component of the group action. The central claim is that this structural constraint increases the effective sample size from $N$ to $d_{\mathrm{iso}}N$, and that for rare-event conditioning the relevant probability denominator shrinks from $\sqrt{P(x\in A)}$ to $\sqrt{|G'|\,P(x\in A)}$. If correct, the result provides the first non-asymptotic statistical learning guarantees for equivariant conditional inference with learned neural features, and it makes uncertainty quantification a byproduct of regression in a single model.

What carries the argument

The load-bearing object is the conditional expectation operator viewed as a G-equivariant linear integral operator between symmetric Hilbert spaces. Its isotypic decomposition $L^2=\bigoplus_{k=1}^{n_{\mathrm{iso}}} L^2_{(k)}$ splits the operator into blocks, and the group action forces each block to have the Kronecker form $\Theta^{(k)}\otimes I_{d_k}$, where $d_k$ is the dimension of the smallest faithful representation of the relevant subgroup; this is the mechanism that injects extra sample efficiency. The symmetry index $\gamma_{G'}(A)$ measures how much the conditioning set overlaps its own orbit, and group-averaged U-statistics provide unbiased estimates of the contrastive loss with effective sample size $|G|N$ and $d_kN^2$.

What would settle it

Run eNCP on a symmetric cGMM where the conditioning set $A$ is neither invariant nor anti-symmetric, so $0<\gamma_{G'}(A)<1$, and compare the empirical error rate against $1/\sqrt{|G'|\,P(x\in A)}$ with the rate $1/\sqrt{P(x\in G'\triangleright A)}$; whichever denominator governs the observed rate settles which rare-event bound is real. A second decisive check is a target whose isotypic multiplicities $m_k$ differ across blocks, since the proof's balancing step assumes they are equal.

Watch

Extended reading notes

Core claim

Under the assumptions that the joint distribution and the marginal of $x$ are G-invariant, the conditional expectation operator $E_{y|x}:L^2_y\to L^2_x$ is G-equivariant and its kernel is the G-invariant pointwise mutual dependency. By the isotypic decomposition of the symmetric Hilbert spaces, the operator disentangles into blocks $E^{(k)}_{y|x}$; each block is parametrized as $\Theta^{(k)}\otimes I_{d_k}$, so its singular spaces have dimension at least $d_k$, the dimension of the corresponding irreducible representation. The paper defines eNCP as the model that learns these blocks with an equivariant bilinear contrastive loss, then reads off regression and conditional probabilities from the same learned operator. Theorem 5.1 states that, with probability at least $1-\delta$, the $L^2$ error of the estimated conditional expectation $\hat z_\theta$ is bounded by $\sqrt{\mathrm{Var}[\|h\|]}$ times $[E^r_\theta + (d_{\mathrm{iso}}N)^{-\alpha/(1+2\alpha)}\log(n_{\mathrm{iso}}/\delta)]$, while the set-conditioned error has the same factor divided by $\sqrt{|G'|\,P(x\in A)}$ and multiplied by a symmetry-index term $\gamma_{G'}(A)$.

Load-bearing premise

The load-bearing premise is unstated in Theorem 5.1 but used in its proof: the conditioning set $A$ must be anti-symmetric, all isotypic multiplicities must be equal, the observable $h$ must be sub-Gaussian, and the symmetry-index bound of Assumption M.6 must hold; if any of these fails, the advertised effective-sample-size and rare-event gains are not proven.

Editorial extensions

If this is right

  • The same trained model yields G-equivariant regression, conditional probabilities, and confidence intervals at any coverage level without retraining, because it regresses the full conditional CDF.
  • Sample complexity of equivariant conditional inference improves from $N^{-\alpha/(1+2\alpha)}$ to $(d_{\mathrm{iso}}N)^{-\alpha/(1+2\alpha)}$, a gain that grows with both the number of isotypic components and the dimensions of the faithful representations.
  • For an asymmetric conditioning set $A$, the rare-event probability denominator is multiplied by $|G'|$, so events with probability $P(x\in A)$ can be conditioned on as if they had probability $|G'|P(x\in A)$.
  • The framework is architecture-agnostic: any G-equivariant backbone can be used, and the guarantees connect the quality of the learned representation $E^r_\theta=\|E_{y|x}-E_\theta\|_{\mathrm{op}}$ directly to the downstream error.
  • Setting $G$ to the trivial group recovers the symmetry-agnostic NCP baseline, so the guarantees include the earlier method as a special case.

Reading between the lines

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

  • A testable consequence not drawn in the paper: because the learned operator is a surrogate for the regression loss, the same machinery should provide a surrogate-gap bound that quantifies how much the contrastive training loss overestimates the equivariant regression MSE.
  • The rare-event result suggests a data-augmentation strategy: for a rare asymmetric event $A$, conditioning on the group orbit of $A$ is statistically as informative as observing roughly $|G'|$ independent samples of the event; this could be verified by comparing eNCP to a model trained with standard orbit averaging.
  • The equal-multiplicity assumption is likely a proof artifact rather than a structural requirement; if true in practice, allowing $m_k$ to differ would let the framework adapt to groups whose irreps appear with very different frequencies in a target.
  • One could turn the framework into a symmetry test: eNCP's validation performance versus its symmetry-agnostic counterpart measures how much of the conditional structure is actually explained by $G$, suggesting a score for detecting symmetry misspecification.
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

4 major / 5 minor

Summary. The manuscript introduces Equivariant Neural Conditional Probability (eNCP), an extension of the Neural Conditional Probability framework of Kostic et al. that incorporates finite group symmetries into representation learning for conditional distributions. The paper argues that, under G-invariance of the joint distribution, the conditional expectation operator decomposes into isotypic components, yielding disentangled equivariant representations and block-diagonal equivariant operator approximations. The central theoretical contribution is Theorem 5.1, which claims non-asymptotic L2 regression bounds with effective sample size amplified by the group structure and a rare-event bound for conditional probabilities controlled by the symmetry index of the conditioning set. The empirical section reports gains over symmetry-agnostic baselines on synthetic conditional Gaussian mixtures, quadruped momentum regression, and uncertainty quantification for ground-reaction forces and mechanical work in legged locomotion.

Significance. The intended contribution is significant: if the theorem is established under the conditions stated in the main text, this would be the first non-asymptotic statistical learning guarantee for equivariant conditional models with neural features, and the rare-event bound would quantify a concrete symmetry-induced variance reduction. The paper is also constructive: it gives a concrete architecture that can wrap any equivariant backbone, proposes a disentangled loss, and ships reproducible experiments with an analytical PMD benchmark that is well suited to isolating approximation error. However, the current version substantially overclaims: the theorem stated in Section 5 is not supported by the appendix proofs without additional assumptions and with a different rate in the regression case. These mismatches are load-bearing because they concern precisely the advertised effective sample size and rare-event improvements. The underlying equivariant operator decomposition and the concentration argument for the symmetry index appear to be a genuine new extension of the prior NCP work, not a circular restatement, but the stated theorem must be corrected before the guarantees can be accepted.

major comments (4)
  1. [§5, Thm 5.1 vs App. M, Thm M.7 and Thm M.8] The statement of Theorem 5.1 claims bounds under only G-invariance, a Schatten-class condition, and an equivariant or invariant h, for any measurable A and any subgroup G′. The appendix proof does not establish this: Theorem M.7 additionally assumes that A is anti-symmetric for G, that mk = m for all k, that h and f are sub-Gaussian, and that Assumption M.6 holds; Theorem M.8 additionally assumes that Y is sub-Gaussian and that Eθ ≤ 1. None of these assumptions appear in Theorem 5.1. This is not a cosmetic mismatch: for G = {e}, y ∼ t3, and h(y) = y, all hypotheses stated in Theorem 5.1 are satisfied, but the sample-mean tail is polynomial, so the threshold implied by (19) is false in that base case. The theorem statement must either include the missing regularity and symmetry assumptions or state weaker bounds that follow from the provided proofs.
  2. [§5, Eq. (20) vs App. M, Thm M.7 and Lemma M.10] The rare-event bound (20) omits the relative-error term ηA that the appendix introduces and uses to derive the advertised orbit-probability gain. Theorem M.7's conditional-probability display contains the additive term sqrt(|G| ηA), and Lemma M.10 only produces the clean orbit-probability denominator after assuming g ▷ A ∩ A = ∅ for all nontrivial g. In the main text, ηA is absent and the anti-symmetry condition is not stated, so the bound in (20) is not the one proved in the appendix. The statement must include the anti-symmetry condition and the ηA term, or the rare-event claim must be weakened correspondingly.
  3. [§5, Thm 5.1 and discussion (item (ii)) vs App. M, Thm M.8] The regression rate in Theorem 5.1 is stated with effective sample size disoN, but the appendix theorem that proves the regression bound, Theorem M.8, yields after balancing the dimension m a rate of the form (diso|G|N)^{-2α/(1+2α)} in squared error, i.e., an effective sample size of diso|G|N, not disoN. The main text's statement that symmetries 'boost the effective sample size to disoN' is therefore not supported by the appendix derivation if the appendix rate is the correct one; if the intended rate is instead disoN, a proof of that rate is missing. This discrepancy affects a central advertised contribution and must be resolved.
  4. [§5, Thm 5.1 vs App. M, Thm M.8] Theorem M.8 also requires the sample-size condition N ≥ (cu ∨ cv)^2 m log(eδ^{-1}niso) ∨ |G| and the condition Eθ ≤ 1, while Theorem 5.1 states 'with probability at least 1 − δ' with no such restrictions. The missing Eθ ≤ 1 condition is not innocuous, since Eθ is an optimization error that the theorem treats as an arbitrary input. Without these conditions, the claimed high-probability statement (19) is not a consequence of the appendix theorem.
minor comments (5)
  1. [§5, Eq. (18) and App. M, Eq. (111)] The same symbol γG′ is used for the symmetry index of a set A and for the symmetry index of an observable h; these are different quantities with different ranges and should be denoted differently to avoid confusion.
  2. [App. I.2] The sentence 'For intution refeer to the following blog post' contains typos and is followed by no reference or URL; please complete or delete this sentence.
  3. [App. B] The entry for eCQR reads 'Version of eCQR where...' and should read 'Version of CQR where...'.
  4. [App. A] The appendix overview says 'App. A summarizes the notations used, while App. A provides a glossary'; the second reference should presumably be to another appendix or should be removed.
  5. [App. G and §6] The code reproducibility statement mentions an open-access repository and package 'symm_rep_learn', but no URL or persistent identifier is provided; please include one.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity; the claimed guarantees are derived from operator-theoretic approximation and concentration arguments, with the representation error E_theta appearing as an assumed input rather than as a fitted prediction.

full rationale

Walking the derivation chain, Theorem 5.1 takes the representation learning error E_r_theta = ||E_y|x - E_theta||_op as an input and proves the L2 and set-conditional bounds via the SVD truncation analysis of Appendix M, group-averaged empirical means, and concentration inequalities (Thm M.2, Lemma M.9, Lemma M.10). The symmetry index gamma_G'(A) is not fitted from data and is not the target of prediction; it is a set-overlap measure that appears naturally through group averaging in the proof. The citations to the authors' prior NCP work [36,37] supply the surrogate-loss inequality E_r_theta <= sqrt(L_gamma(theta)-L_gamma(star)); that is a parameter-free mathematical result from earlier papers, not a constructional equivalence, and the new equivariant rates do not reduce to it by definition. The main correctness concern is that the appendix proof of Theorems M.7 and M.8 assumes conditions absent from the statement of Theorem 5.1, including anti-symmetric A, equal multiplicities m_k = m, sub-Gaussianity of h and Y, and Assumption M.6; this is a proof gap or overclaim, but it is not circularity. No fitted parameter is renamed as a prediction, no load-bearing self-citation forbids alternatives, and no ansatz is smuggled in via citation. The central new derivations are self-contained relative to the paper's assumptions, so the circularity score is 0.

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

The central guarantees rest on exact symmetry, Schatten-class decay, sub-Gaussianity, and two appendix-only technical assumptions. No new physical entities are introduced. The E_theta blocks are trained model parameters, and the balancing choice for m is an asymptotic order, not a fitted constant.

assumptions (7)
  • domain assumption P(y|x) and P(x) are exactly G-invariant under a finite symmetry group G (Eq. 6)
    This is the fundamental prior of the paper. If it is misspecified, the guarantees collapse, and the authors list this as a limitation.
  • standard math The joint distribution is absolutely continuous with respect to the product of marginals, so the PMD kappa and conditional expectation operator are well defined
    Needed to define E_y|x via a kernel in App. H and to justify the operator-theoretic formulation.
  • domain assumption E_y|x is a compact (1/alpha)-Schatten class operator with singular values decaying as sigma_i <= c i^(-alpha) (Assumption M.5)
    Controls the bias and approximation error and sets the rate exponent alpha/(1+2alpha). This decay is not derived from symmetry or data.
  • ad hoc to paper The observable h and the response y are sub-Gaussian
    Used in concentration lemmas (M.9, M.14) but not stated in Thm 5.1, so the main theorem is stronger than the proof supports.
  • ad hoc to paper Assumption M.6: (|G'|-1) gamma_G'(h) <= C0 E[||h||^2]
    A technical control on the autocorrelation of h under G' that is absent from the main theorem statement.
  • ad hoc to paper The conditioning event A is anti-symmetric for G and m_k = m for all k
    Added only in the Thm M.7 proof and not in Thm 5.1, so the general measurable-A result is not proven as stated.
  • standard math Isotypic decomposition of G-symmetric Hilbert spaces (Thm I.8) and Schur's lemma (Lem I.12)
    Core structural facts used to block-diagonalize the operator and constrain the SVD to singular spaces of dimension at least d_k.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Representation Learning for Equivariant Inference with Guarantees." pith.science (2026). https://pith.science/paper/AI5WRDJ5

@misc{pith2026250519809,
  author       = {Pith},
  title        = {Pith review of: Representation Learning for Equivariant Inference with Guarantees},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/AI5WRDJ5}},
  note         = {Machine review of arXiv:2505.19809}
}
read the original abstract

In many real-world applications of regression, conditional probability estimation, and uncertainty quantification, exploiting symmetries rooted in physics or geometry can dramatically improve generalization and sample efficiency. While geometric deep learning has made empirical advances by incorporating symmetry and geometry priors, less attention has been given to statistical learning guarantees. In this paper, we introduce an equivariant representation learning framework that simultaneously addresses regression, conditional probability estimation, and uncertainty quantification while providing first-of-its-kind non-asymptotic statistical learning guarantees. Grounded in operator and group representation theory, our framework approximates the spectral decomposition of the conditional expectation operator, building representations that are both equivariant and disentangled along independent symmetry quotient groups. Empirical evaluations on synthetic datasets and real-world robotics applications confirm the potential of our approach, matching or outperforming existing equivariant baselines in regression while providing well-calibrated uncertainty estimates.

Figures

Figures reproduced from arXiv: 2505.19809 by the authors.

Figure 1
Figure 1. Left: Test set sample efficiency for G-equivariant regression (MSE vs. training samples) when predicting the G-equivariant linear and angular momentum of a quadruped robot’s center of mass (CoM) from noisy joint positions and velocities. Right: Uncertainty quantification via G-equivariant prediction of 90% confidence intervals (CI, light-red area) for the robot’s instantaneous work Ut and kinetic energy Tt during lo… view at source ↗
Figure 2
Figure 2. Left: NCP’s bilinear NN architecture. Right: eNCP’s G-equivariant bilinear NN architecture, featuring ϕθ and ψθ as G-equivariant NNs and Eθ as a G-equivariant block-diagonal matrix. Each block is equivariant to a subgroup G(k) ≤ G and is constrained to have singular spaces of dimension at least dk—the minimal dimension for a faithful representation of the action of G(k) . and equivariant linear operators (App. K). F… view at source ↗
Figure 3
Figure 3. Example of symmetric random variables (x, y) ∼ X × Y ⊂ R × R, whose marginals P(x) and P(y); joint P(x, y); and conditional P(y|x) distributions are invariant to reflections of the data: gr ▷X x = −x and gr ▷Y y = −y, where gr denotes the reflection element of the reflection symmetry group C2 := {e, gr|g 2 r = e}. Consequently, the PMD κ(x, y) is C2-invariant. Consequently, NCP parameterizes Eθ by a bilinear model κ… view at source ↗
Figures from the paper (10 more)
Figure 4
Figure 4. Figure 4: Sample efficiency plots comparing the test set [PITH_FULL_IMAGE:figures/full_fig_p008_4.png]
Figure 5
Figure 5. Figure 5: Sample efficiency plots comparing test set regression mean-square-error of the density [PITH_FULL_IMAGE:figures/full_fig_p022_5.png]
Figure 6
Figure 6. Figure 6: Example of morphological finite symmetry in robotics. [PITH_FULL_IMAGE:figures/full_fig_p023_6.png]
Figure 7
Figure 7. Figure 7: Synthetic experiment in uncertainty quantification, originally proposed by Feldman et al. [PITH_FULL_IMAGE:figures/full_fig_p024_7.png]
Figure 8
Figure 8. Figure 8: Prediction of the 80% and 95% confidence intervals for the random variable y in experiment App. G.3 using the proposed eNCP model. The model estimates the CCDF by discretizing each dimension of y = [y1 , y2 ] into 100 bins and computing the conditional probabilities P(…
Figure 9
Figure 9. Figure 9: Left: Learned basis functions from the NCP model for y = [y0 , y1 ]. Right: Learned basis functions from the eNCP model for y. The marginal distribution of y exhibits reflection symmetry gr ▷Y y = [−y0 , y1 ] under G = C2. Incorporating this prior, the eNCP model decom…
Figure 10
Figure 10. Figure 10: Results of a synthetic experiment in uncertainty quantification comparing [PITH_FULL_IMAGE:figures/full_fig_p026_10.png]
Figure 11
Figure 11. Figure 11: Prediction of 90% confidence intervals (CI) for the ground-reaction forces [PITH_FULL_IMAGE:figures/full_fig_p028_11.png]
Figure 12
Figure 12. Figure 12: Left: Diagram of the group action ▷F on functions f1(x) = x 2+c and f2(x) = x 3 defined on the domain X := R endowed with the reflectional symmetry group G := C2 = {e, gs}, with the reflection action acting on the domain by gs ▷ x = −x and on the function space F := {…
Figure 13
Figure 13. Figure 13: Visualization of the basis functions in the finite-dimensional symmetric function space [PITH_FULL_IMAGE:figures/full_fig_p036_13.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

87 extracted references · 61 canonical work pages

  1. [1]

    Joint measures and cross-covariance operators

    Charles R Baker. Joint measures and cross-covariance operators. Trans. Am. Math. Soc., 186: 273–273, 1973

    1973

  2. [2]

    On the surrogate gap between contrastive and supervised losses

    Han Bao, Yoshihiro Nagano, and Kento Nozawa. On the surrogate gap between contrastive and supervised losses. In International conference on machine learning, pages 1585–1606. PMLR, 2022

    2022

  3. [3]

    Representation learning: A review and new perspectives

    Yoshua Bengio, Aaron Courville, and Pascal Vincent. Representation learning: A review and new perspectives. IEEE transactions on pattern analysis and machine intelligence , 35(8): 1798–1828, 2013

    2013

  4. [4]

    Concentration Inequalities for Sums and Martingales

    Bernard Bercu, Bernard Delyon, and Emmanuel Rio. Concentration Inequalities for Sums and Martingales. SpringerBriefs in Mathematics. Springer, 2015

    2015

  5. [5]

    On the sample complexity of learning under geometric stability

    Alberto Bietti, Luca Venturi, and Joan Bruna. On the sample complexity of learning under geometric stability. Advances in neural information processing systems, 34:18673–18684, 2021

    2021

  6. [6]

    Contact model fusion for event-based locomotion in unstructured terrains

    Gerardo Bledt, Patrick M Wensing, Sam Ingersoll, and Sangbae Kim. Contact model fusion for event-based locomotion in unstructured terrains. In 2018 IEEE International Conference on Robotics and Automation (ICRA), pages 4399–4406. IEEE, 2018

    2018

  7. [7]

    Clifford neural layers for pde modeling

    Johannes Brandstetter, Rianne van den Berg, Max Welling, and Jayesh K Gupta. Clifford neural layers for pde modeling. arXiv preprint arXiv:2209.04934, 2022

    arXiv 2022

  8. [8]

    Geometric deep learning: Grids, groups, graphs, geodesics, and gauges

    Michael M Bronstein, Joan Bruna, Taco Cohen, and Petar Veliˇckovi´c. Geometric deep learning: Grids, groups, graphs, geodesics, and gauges. arXiv preprint arXiv:2104.13478, 2021

    arXiv 2021

Show all 87 references
  1. [9]

    La théorie des groupes finis et continus et l’analysis situs

    Élie Cartan. La théorie des groupes finis et continus et l’analysis situs. Number 42 in Mémorial des sciences mathématiques. Gauthier-Villars, 1952

    1952

  2. [10]

    A simple framework for contrastive learning of visual representations

    Ting Chen, Simon Kornblith, Mohammad Norouzi, and Geoffrey Hinton. A simple framework for contrastive learning of visual representations. In International conference on machine learning, pages 1597–1607. PmLR, 2020

    2020

  3. [11]

    When does contrastive visual representation learning work? In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 14755–14764, June 2022

    Elijah Cole, Xuan Yang, Kimberly Wilber, Oisin Mac Aodha, and Serge Belongie. When does contrastive visual representation learning work? In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 14755–14764, June 2022

    2022

  4. [12]

    Equivariant self-supervised learning: Encouraging equivariance in representations

    Rumen Dangovski, Li Jing, Charlotte Loh, Seungwook Han, Akash Srivastava, Brian Che- ung, Pulkit Agrawal, and Marin Soljacic. Equivariant self-supervised learning: Encouraging equivariance in representations. In International Conference on Learning Representations

  5. [13]

    Equivariant contrastive learning

    Rumen Dangovski, Li Jing, Charlotte Loh, Seungwook Han, Akash Srivastava, Brian Che- ung, Pulkit Agrawal, and Marin Solja ˇci´c. Equivariant contrastive learning. In International Conference on Learning Representations, 2022

    2022

  6. [14]

    How rotational invariance of common kernels prevents generalization in high dimensions

    Konstantin Donhauser, Mingqi Wu, and Fanny Yang. How rotational invariance of common kernels prevents generalization in high dimensions. In International Conference on Machine Learning, pages 2804–2814. PMLR, 2021

    2021

  7. [15]

    Group theory: application to the physics of condensed matter

    Mildred S Dresselhaus, Gene Dresselhaus, and Ado Jorio. Group theory: application to the physics of condensed matter. Springer Science & Business Media, 2007

    2007

  8. [16]

    The approximation of one matrix by another of lower rank

    Carl Eckart and Gale Young. The approximation of one matrix by another of lower rank. Psychometrika, 1(3):211–218, 1936

    1936

  9. [17]

    Symmetry and generalisation in machine learning

    B Elesedy. Symmetry and generalisation in machine learning. PhD thesis, University of Oxford, 2023

    2023

  10. [18]

    Provably strict generalisation benefit for invariance in kernel methods

    Bryn Elesedy. Provably strict generalisation benefit for invariance in kernel methods. In Advances in Neural Information Processing Systems, volume 34, pages 17273–17283. Curran Associates, Inc., 2021. URL https://proceedings.neurips.cc/paper_files/paper/ 2021/file/8fe04df45a2...

    2021

  11. [19]

    Provably strict generalisation benefit for equivariant models

    Bryn Elesedy and Sheheryar Zaidi. Provably strict generalisation benefit for equivariant models. In Marina Meila and Tong Zhang, editors, Proceedings of the 38th International Conference on Machine Learning, volume 139 of Proceedings of Machine Learning Research, pages 2959–29...

    2021

  12. [20]

    Calibrated multiple-output quantile regression with representation learning

    Shai Feldman, Stephen Bates, and Yaniv Romano. Calibrated multiple-output quantile regression with representation learning. Journal of Machine Learning Research, 24(24):1–48, 2023

    2023

  13. [21]

    Dimensionality reduction for supervised learning with reproducing kernel hilbert spaces

    Kenji Fukumizu, Francis R Bach, and Michael I Jordan. Dimensionality reduction for supervised learning with reproducing kernel hilbert spaces. Journal of Machine Learning Research, 5(Jan): 73–99, 2004

    2004

  14. [22]

    Dynamic identification of robots with power model

    Maxime Gautier. Dynamic identification of robots with power model. In Proceedings of international conference on robotics and automation, volume 3, pages 1922–1927. IEEE, 1997

    1922

  15. [23]

    Conditional gaussian mixture models for environmental risk mapping

    Nicolas Gilardi, Samy Bengio, and Mikhail Kanevski. Conditional gaussian mixture models for environmental risk mapping. In Proceedings of the 12th IEEE workshop on neural networks for signal processing, pages 777–786. IEEE, 2002

    2002

  16. [24]

    Springer Science & Business Media, 2012

    Martin Golubitsky, Ian Stewart, and David G Schaeffer.Singularities and Groups in Bifurcation Theory: Volume II, volume 69. Springer Science & Business Media, 2012

    2012

  17. [25]

    Structuring representation geometry with rotationally equivariant contrastive learning

    Sharut Gupta, Joshua Robinson, Derek Lim, Soledad Villar, and Stefanie Jegelka. Structuring representation geometry with rotationally equivariant contrastive learning. In The Twelfth International Conference on Learning Representations, 2023

    2023

  18. [26]

    HaoChen, Colin Wei, Adrien Gaidon, and Tengyu Ma

    Jeff Z. HaoChen, Colin Wei, Adrien Gaidon, and Tengyu Ma. Provable guarantees for self-supervised deep learning with spectral contrastive loss. In Advances in Neu- ral Information Processing Systems , volume 34, pages 5000–5011. Curran Associates, Inc., 2021. URL https://proce...

    2021

  19. [27]

    Beyond separability: Analyzing the linear transferability of contrastive representations to related subpopulations

    Jeff Z HaoChen, Colin Wei, Ananya Kumar, and Tengyu Ma. Beyond separability: Analyzing the linear transferability of contrastive representations to related subpopulations. Advances in neural information processing systems, 35:26889–26902, 2022

    2022

  20. [28]

    Data-efficient image recognition with contrastive predictive coding

    Olivier Henaff. Data-efficient image recognition with contrastive predictive coding. In Interna- tional conference on machine learning, pages 4182–4192. PMLR, 2020

    2020

  21. [29]

    Towards a definition of disentangled representations

    Irina Higgins, David Amos, David Pfau, Sebastien Racaniere, Loic Matthey, Danilo Rezende, and Alexander Lerchner. Towards a definition of disentangled representations. arXiv preprint arXiv:1812.02230, 2018

    2018 arXiv

  22. [30]

    Machine Learning Beyond Point Predictions: Uncertainty Quantification

    Rafael Izbicki. Machine Learning Beyond Point Predictions: Uncertainty Quantification . imprint, 1st edition, 2025. ISBN 978-65-01-20272-3

    2025

  23. [31]

    Rafael Izbicki and Ann B. Lee. Converting high-dimensional regression to high-dimensional conditional density estimation. 2017

    2017

  24. [32]

    Contrastive learning can find an optimal basis for approximately view-invariant functions

    Daniel D Johnson, Ayoub El Hanchi, and Chris J Maddison. Contrastive learning can find an optimal basis for approximately view-invariant functions. arXiv preprint arXiv:2210.01883, 2022

    2022 arXiv

  25. [33]

    Physics-informed machine learning: case studies for weather and climate modelling

    Karthik Kashinath, M Mustafa, Adrian Albert, JL Wu, C Jiang, Soheil Esmaeilzadeh, Kamyar Azizzadenesheli, R Wang, Ashesh Chattopadhyay, A Singh, et al. Physics-informed machine learning: case studies for weather and climate modelling. Philosophical Transactions of the Royal So...

    2021

  26. [34]

    Homomorphism autoencoder–learning group structured representations from observed transi- tions

    Hamza Keurti, Hsiao-Ru Pan, Michel Besserve, Benjamin F Grewe, and Bernhard Schölkopf. Homomorphism autoencoder–learning group structured representations from observed transi- tions. In International Conference on Machine Learning, pages 16190–16215. PMLR, 2023. 11

    2023

  27. [35]

    Anthony W. Knapp. Representation Theory of Semisimple Groups, An Overview Based on Examples (PMS-36). Princeton University Press, Princeton, 1986

    1986

  28. [36]

    Learning invariant representations of time-homogeneous stochastic dynamical systems

    Vladimir R Kostic, Pietro Novelli, Riccardo Grazzi, Karim Lounici, and Massimiliano Pontil. Learning invariant representations of time-homogeneous stochastic dynamical systems. In The Twelfth International Conference on Learning Representations, 2024

    2024

  29. [37]

    Neural conditional probability for uncertainty quantification

    Vladimir R Kostic, Gregoire Pacreau, Giacomo Turri, Pietro Novelli, Karim Lounici, and Massimiliano Pontil. Neural conditional probability for uncertainty quantification. In The Thirty-eighth Annual Conference on Neural Information Processing Systems, 2024

    2024

  30. [38]

    Neural operator: Learning maps between function spaces with applications to pdes

    Nikola Kovachki, Zongyi Li, Burigede Liu, Kamyar Azizzadenesheli, Kaushik Bhattacharya, Andrew Stuart, and Anima Anandkumar. Neural operator: Learning maps between function spaces with applications to pdes. Journal of Machine Learning Research, 24(89):1–97, 2023

    2023

  31. [39]

    Motion graphs

    Lucas Kovar, Michael Gleicher, and Frédéric Pighin. Motion graphs. ACM Trans. Graph., 21 (3):473–482, July 2002. URL https://doi.org/10.1145/566654.566605

    2002

  32. [40]

    Contrastive representation learning: A framework and review

    Phuc H Le-Khac, Graham Healy, and Alan F Smeaton. Contrastive representation learning: A framework and review. Ieee Access, 8:193907–193934, 2020

    2020

  33. [41]

    Mutual information driven equivariant contrastive learning for 3d action representation learning

    Lilang Lin, Jiahang Zhang, and Jiaying Liu. Mutual information driven equivariant contrastive learning for 3d action representation learning. IEEE Transactions on Image Processing, 2024

    2024

  34. [42]

    Hierarchical spacetime control

    Zicheng Liu, Steven J Gortler, and Michael F Cohen. Hierarchical spacetime control. In Proceedings of the 21st annual conference on Computer graphics and interactive techniques, pages 35–42, 1994

    1994

  35. [43]

    George W. Mackey. Harmonic analysis as the exploitation of symmetry–a historical survey. Bulletin (New Series) of the American Mathematical Society, 3(1.P1):543 – 698, 1980

    1980

  36. [44]

    Probabilistic contact state estimation for legged robots using inertial information

    Michael Maravgakis, Despina-Ekaterini Argiropoulos, Stylianos Piperakis, and Panos Trahanias. Probabilistic contact state estimation for legged robots using inertial information. In 2023 IEEE International Conference on Robotics and Automation (ICRA) , pages 12163–12169. IEEE, 2023

    2023

  37. [45]

    Equivariant representation learning via class-pose decomposition

    Giovanni Luca Marchetti, Gustaf Tegnér, Anastasiia Varava, and Danica Kragic. Equivariant representation learning via class-pose decomposition. In International Conference on Artificial Intelligence and Statistics, pages 4745–4756. PMLR, 2023

    2023

  38. [46]

    Learning with invariances in random features and kernel models

    Song Mei, Theodor Misiakiewicz, and Andrea Montanari. Learning with invariances in random features and kernel models. In Conference on Learning Theory, pages 3351–3418. PMLR, 2021

    2021

  39. [47]

    Mitra, Mark Pauly, Michael Wand, and Duygu Ceylan

    Niloy J. Mitra, Mark Pauly, Michael Wand, and Duygu Ceylan. Symmetry in 3d geometry: Extraction and applications. Computer Graphics Forum, 32(6):1–23, February 2013. URL http://dx.doi.org/10.1111/cgf.12010

    2013 doi

  40. [48]

    Learning with group invariant features: A kernel perspective

    Youssef Mroueh, Stephen V oinea, and Tomaso A Poggio. Learning with group invariant features: A kernel perspective. Advances in neural information processing systems, 28, 2015

    2015

  41. [49]

    Evading the curse of dimensionality in nonparametric density estimation with simplified vine copulas

    Thomas Nagler and Claudia Czado. Evading the curse of dimensionality in nonparametric density estimation with simplified vine copulas. Journal of Multivariate Analysis, 151:69–89, October 2016. URL http://dx.doi.org/10.1016/j.jmva.2016.07.003

    2016 doi

  42. [50]

    Muse: A real-time multi-sensor state estimator for quadruped robots

    Ylenia Nisticò, João Carlos Virgolino Soares, Lorenzo Amatucci, Geoff Fink, and Claudio Semini. Muse: A real-time multi-sensor state estimator for quadruped robots. IEEE Robotics and Automation Letters, 2025

    2025

  43. [51]

    Applications of Lie groups to differential equations, volume 107

    Peter J Olver. Applications of Lie groups to differential equations, volume 107. Springer Science & Business Media, 1993

    1993

  44. [52]

    Representation learning with contrastive predictive coding

    Aaron van den Oord, Yazhe Li, and Oriol Vinyals. Representation learning with contrastive predictive coding. arXiv preprint arXiv:1807.03748, 2018. 12

    2018 arXiv

  45. [53]

    Dynamics harmonic analysis of robotic systems: Appli- cation in data-driven koopman modelling

    Daniel Ordoñez-Apraez, Vladimir Kostic, Giulio Turrisi, Pietro Novelli, Carlos Mastalli, Clau- dio Semini, and Massimilano Pontil. Dynamics harmonic analysis of robotic systems: Appli- cation in data-driven koopman modelling. In 6th Annual Learning for Dynamics & Control Confe...

    2024

  46. [54]

    Mor- phological symmetries in robotics

    Daniel Ordoñez-Apraez, Giulio Turrisi, Vladimir Kostic, Mario Martin, Antonio Agudo, Francesc Moreno-Noguer, Massimiliano Pontil, Claudio Semini, and Carlos Mastalli. Mor- phological symmetries in robotics. The International Journal of Robotics Research , 0(0): 027836492412824...

    doi

  47. [55]

    Wasserstein dependency measure for representation learning

    Sherjil Ozair, Corey Lynch, Yoshua Bengio, Aaron van den Oord, Sergey Levine, and Pierre Sermanet. Wasserstein dependency measure for representation learning. In H. Wal- lach, H. Larochelle, A. Beygelzimer, F. d 'Alché-Buc, E. Fox, and R. Garnett, edi- tors, Advances in Neural...

    2019

  48. [56]

    Max-margin invariant features from transformed unlabelled data

    Dipan Pal, Ashwin Kannan, Gautam Arakalgud, and Marios Savvides. Max-margin invariant features from transformed unlabelled data. In Advances in Neural Information Processing Systems, volume 30. Curran Associates, Inc., 2017

    2017

  49. [57]

    Koopman operator and its approximations for systems with symmetries

    Anastasiya Salova, Jeffrey Emenheiser, Adam Rupe, James P Crutchfield, and Raissa M D’Souza. Koopman operator and its approximations for systems with symmetries. Chaos: An Interdisciplinary Journal of Nonlinear Science, 29(9), 2019

    2019

  50. [58]

    DA VID W. Scott. Feasibility of multivariate density estimates. Biometrika, 78(1):197–205,

  51. [59]

    Group symmetry and covariance regularization

    Parikshit Shah and Venkat Chandrasekaran. Group symmetry and covariance regularization. In 2012 46th Annual Conference on Information Sciences and Systems (CISS), pages 1–6. IEEE, 2012

    2012

  52. [60]

    Ralph C. Smith. Uncertainty Quantification: Theory, Implementation, and Applications. Society for Industrial and Applied Mathematics, January 2013. URL http://dx.doi.org/10.1137/ 1.9781611973228

    2013

  53. [61]

    Hilbert space embeddings of conditional distributions with applications to dynamical systems

    Le Song, Jonathan Huang, Alex Smola, and Kenji Fukumizu. Hilbert space embeddings of conditional distributions with applications to dynamical systems. In Proceedings of the 26th Annual International Conference on Machine Learning, pages 961–968, 2009

    2009

  54. [62]

    Density ratio estimation in machine learning

    Masashi Sugiyama, Taiji Suzuki, and Takafumi Kanamori. Density ratio estimation in machine learning. Cambridge University Press, 2012

    2012

  55. [63]

    The exact sample complexity gain from invariances for kernel regression

    Behrooz Tahmasebi and Stefanie Jegelka. The exact sample complexity gain from invariances for kernel regression. Advances in Neural Information Processing Systems, 36, 2023

    2023

  56. [64]

    Contrastive learning, multi-view redundancy, and linear models

    Christopher Tosh, Akshay Krishnamurthy, and Daniel Hsu. Contrastive learning, multi-view redundancy, and linear models. In Proceedings of the 32nd International Conference on Algorithmic Learning Theory, volume 132 of Proceedings of Machine Learning Research, pages 1179–1206. ...

    2021

  57. [65]

    Neural methods for point-wise dependency estimation

    Yao-Hung Hubert Tsai, Han Zhao, Makoto Yamada, Louis-Philippe Morency, and Russ R Salakhutdinov. Neural methods for point-wise dependency estimation. In Advances in Neural Information Processing Systems, volume 33, pages 62–72. Curran Associates, Inc., 2020

    2020

  58. [66]

    Self-supervised representation learning with relative predictive coding

    Yao-Hung Hubert Tsai, Martin Q Ma, Muqiao Yang, Han Zhao, Louis-Philippe Morency, and Ruslan Salakhutdinov. Self-supervised representation learning with relative predictive coding. In International Conference on Learning Representations, 2021

    2021

  59. [67]

    On the benefits of gpu sample-based stochastic predictive controllers for legged locomotion

    Giulio Turrisi, Valerio Modugno, Lorenzo Amatucci, Dimitrios Kanoulas, and Claudio Semini. On the benefits of gpu sample-based stochastic predictive controllers for legged locomotion. In 2024 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), pages 137...

    2024

  60. [68]

    Mdp homomorphic networks: Group symmetries in reinforcement learning

    Elise van der Pol, Daniel Worrall, Herke van Hoof, Frans Oliehoek, and Max Welling. Mdp homomorphic networks: Group symmetries in reinforcement learning. In Advances in Neural Information Processing Systems, volume 33, pages 4199–4210. Curran Associates, Inc., 2020

  61. [69]

    Introduction to the non-asymptotic analysis of random matrices

    Roman Vershynin. Introduction to the non-asymptotic analysis of random matrices. arXiv:1011.3027, 2011

    2011 arXiv

  62. [70]

    Towards understanding the mechanism of contrastive learning via similarity structure: A theoretical analysis

    Hiroki Waida, Yuichiro Wada, Léo Andéol, Takumi Nakagawa, Yuhui Zhang, and Takafumi Kanamori. Towards understanding the mechanism of contrastive learning via similarity structure: A theoretical analysis. In Joint European Conference on Machine Learning and Knowledge Discovery ...

    2023

  63. [71]

    Incorporating symmetry into deep dynamics models for improved generalization

    R Wang. Incorporating symmetry into deep dynamics models for improved generalization. In International Conference on Learning Representations (ICLR), 2021

    2021

  64. [72]

    Incorporating symmetry into deep dynamics models for improved generalization

    Rui Wang, Robin Walters, and Rose Yu. Incorporating symmetry into deep dynamics models for improved generalization. arXiv preprint arXiv:2002.03061, 2020

    2002 arXiv

  65. [73]

    Approximately equivariant networks for imperfectly symmetric dynamics

    Rui Wang, Robin Walters, and Rose Yu. Approximately equivariant networks for imperfectly symmetric dynamics. In International Conference on Machine Learning, pages 23078–23091. PMLR, 2022

    2022

  66. [74]

    Disentangled representation learning

    Xin Wang, Hong Chen, Si’ao Tang, Zihao Wu, and Wenwu Zhu. Disentangled representation learning. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2024

    2024

  67. [75]

    Understand- ing the role of equivariance in self-supervised learning

    Yifei Wang, Kaiwen Hu, Sharut Gupta, Ziyu Ye, Yisen Wang, and Stefanie Jegelka. Understand- ing the role of equivariance in self-supervised learning. In The Thirty-eighth Annual Conference on Neural Information Processing Systems, 2024

    2024

  68. [76]

    Spectral representation learning for conditional moment models

    Ziyu Wang, Yucen Luo, Yueru Li, Jun Zhu, and Bernhard Schölkopf. Spectral representation learning for conditional moment models. arXiv preprint arXiv:2210.16525, 2022

    2022 arXiv

  69. [77]

    All of Nonparametric Statistics

    Larry Wasserman. All of Nonparametric Statistics. Springer Texts in Statistics. Springer, New York, NY , 1 edition, May 2007

    2007

  70. [78]

    World Scientific, 2023

    Maurice Weiler, Patrick Forré, Erik Verlinde, and Max Welling.Equivariant and Coordinate In- dependent Convolutional Networks. World Scientific, 2023. URL https://maurice-weiler. gitlab.io/cnn_book/EquivariantAndCoordinateIndependentCNNs.pdf

    2023

  71. [79]

    Latent space symmetry discovery

    Jianke Yang, Nima Dehmamy, Robin Walters, and Rose Yu. Latent space symmetry discovery. In International Conference on Machine Learning, 2023

    2023

  72. [80]

    Contrastive-equivariant self-supervised learning improves alignment with primate visual area it

    Thomas Yerxa, Jenelle Feather, Eero Simoncelli, and SueYeon Chung. Contrastive-equivariant self-supervised learning improves alignment with primate visual area it. Advances in neural information processing systems, 37:96045–96070, 2024

    2024

  73. [81]

    Sample Efficient Grasp Learning Using Equivariant Models

    Xupeng Zhu, Dian Wang, Ondrej Biza, Guanang Su, Robin Walters, and Robert Platt. Sample Efficient Grasp Learning Using Equivariant Models. In Proceedings of Robotics: Science and Systems, New York City, NY , USA, June 2022

    2022

  74. [82]

    Contrastive learning using spectral methods

    James Y Zou, Daniel J Hsu, David C Parkes, and Ryan P Adams. Contrastive learning using spectral methods. In Advances in Neural Information Processing Systems, volume 26. Curran Associates, Inc., 2013. 14 Appendix Table of Contents A Symbols and notation 16 B Acronyms 17 C Rel...

    2013

  75. [84]

    (see Fig. 3). This kernel function enable us to express the joint probability by: P(x ∈ A, y ∈ B)= Z X ×Y 1A(x)1B(y) κ(x, y)Py(dy)Px(dx)| {z } Pxy (dx,dy) = Z A×B k(x, y)Px(dx)Py(dy). (34) Furthermore, given that P(y∈B|x∈A) = P(x∈A, y∈B)/P(x ∈ A), the conditional probability o...

  76. [85]

    Tails: P{|Z| > t} ≤exp 1 − t2/K 2 1 for all t ≥ 0

  77. [86]

    Moments: (E|Z|p)1/p ≤ K2 √p for all p ≥ 1

  78. [87]

    A random variable Z satisfying any of the above assertions is called a sub-Gaussian random variable

    Super-exponential moment: E exp Z 2/K 2 3 ≤ 2. A random variable Z satisfying any of the above assertions is called a sub-Gaussian random variable. We will denote by K3 the sub-Gaussian norm. Consequently, a sub-Gaussian random variable satisfies the following equivalence of m...

  79. [1991]

    URL http://dx.doi.org/10.1093/biomet/78.1.197

    doi

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.