Pith. sign in

REVIEW 2 major objections 4 minor 68 references

Algebra Unveils Deep Learning -- An Invitation to Neuroalgebraic Geometry

T0 review · 2 major / 4 minor · reviewed 2026-08-09 · deepseek-v4-flash

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

Pith's one-line read The paper claims that function spaces of polynomial-activation networks are semi-algebraic varieties, and that their algebro-geometric invariants control core aspects of learning.

desk verdict A useful, well-written position paper proposing 'neuroalgebraic geometry' as a research program; the main weakness is an overreaching approximation claim in Section 5.1. read the letter →

arxiv 2501.18915 v2 pith:KSTXNQSM submitted 2025-01-31 cs.LG math.AG

Giovanni Luca Marchetti , Vahid Shahverdi , Stefano Mereta , Matthew Trager , Kathlén Kohn This is my paper · ORCID
classification cs.LGmath.AG MSC 14P1068T0762R01
keywords neuroalgebraicgeometryneuromanifoldsemi-algebraicvarietyalgebraicsamplecomplexityimplicitbiassingularitiesmetric
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper proposes that the function spaces parameterized by machine learning models—the neuromanifolds—should be studied with algebraic geometry rather than differential geometry alone. Its central claim is that for algebraic models, meaning networks with polynomial or piecewise-polynomial activation functions, the neuromanifold is a semi-algebraic variety. It then argues that algebro-geometric invariants of that variety, such as dimension, degree, singularities, fibers, and critical points, control sample complexity, expressivity, training dynamics, and implicit bias. If the claim is right, a precise dictionary becomes available: architecture choices translate into computable geometric invariants, and learning becomes a distance problem over these varieties. The paper offers this dictionary as an invitation to a research direction it calls neuroalgebraic geometry.

What carries the argument

The central object is the neuromanifold $\mathcal{M}=\{f_w : w\in\mathcal{W}\}$, the image of the parameterization map $\varphi:\mathcal{W}\to\mathcal{V}$ inside a finite-dimensional ambient space of functions $\mathcal{V}$. For algebraic models $\varphi$ is polynomial, so $\mathcal{M}$ is a semi-algebraic variety, meaning a set carved out by polynomial equalities and inequalities. The machinery is the pair $(\mathcal{M},\varphi)$: intrinsic invariants of the variety (dimension, degree, singularities) plus the map's fibers and critical points. Metric algebraic geometry then supplies the distance-based tools—covering-number bounds from dimension and degree, Voronoi cells around singularities, Euclidean distance degree, and data discriminants—that turn learning questions into concrete geometric computations.

What would settle it

Fix a small MLP architecture with a smooth non-polynomial activation, take compact parameter and input domains, and compute the Hausdorff distance between the true neuromanifold and the neuromanifold obtained by replacing the activation with its degree-$d$ polynomial approximation; if this distance does not go to zero as $d$ grows, the paper's bridge from algebraic models to continuous networks breaks.

Watch

Extended reading notes

Core claim

For a parametric model whose parameterization map $\varphi: \mathcal{W} \to \mathcal{V}$ is polynomial in both parameters and inputs, the image of $\varphi$—the neuromanifold $\mathcal{M}=\{f_w : w\in\mathcal{W}\}$—is a semi-algebraic variety by the Tarski–Seidenberg theorem. The paper argues that the invariants of this variety are machine-learning-relevant: dimension and degree bound covering numbers and hence sample complexity; singularities correspond to subnetworks and create implicit bias; fibers of the parameterization encode identifiability and symmetries; critical points and data discriminants structure the loss landscape; and Euclidean distance degree quantifies the complexity of distance minimization, and thus of fitting. Together these claims amount to a dictionary in which algebraic geometry is not an analogy but a working formalism for deep learning theory.

Load-bearing premise

The load-bearing premise is that approximating each activation function by polynomials on a compact set also approximates the entire function space closely enough in Hausdorff distance—a step the paper sketches but does not prove.

Editorial extensions

If this is right

  • Sample complexity is governed by dimension and degree: the paper cites bounds of the form $\log N_\varepsilon(\mathcal{M}) = O\!\left(m \log\frac{d}{\varepsilon} + C\right)$, converting covering-number bounds into sample-size guarantees.
  • Expressivity is quantified as a tubular volume: the set of functions within distance $\varepsilon$ of the neuromanifold has volume bounded by the covering number, so a model's approximate expressive power is controlled by its dimension and degree.
  • Singular points of the neuromanifold act as implicit biases and, in many architectures, correspond exactly to subnetworks, offering a geometric explanation for automatic selection of simpler functions.
  • The loss landscape is organized by critical-point invariants: spurious critical points come from the parameterization's critical locus, and the number and type of real critical points change only across data discriminants.
  • The framework extends beyond polynomial activations: continuous activations can be approximated on compact sets by polynomials, and ReLU networks can be treated with tropical geometry, so the algebraic results carry over approximately.

Reading between the lines

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

  • If the dictionary is correct, architecture design becomes prescriptive: one could choose depth, width, and activation degree to target a desired dimension, degree, or singularity locus of the neuromanifold.
  • The Hausdorff-approximation step in Section 5.1, once made fully rigorous, would transfer singularity and critical-point results to smooth practical activation functions; the paper sketches but does not prove this transfer.
  • A testable extension is to check, in a fixed small architecture, whether the number and type of spurious critical points predicted by Euclidean distance degree matches gradient descent's behavior on synthetic data with known ground truth.
  • The proposed dictionary also suggests a unification of singular learning theory and algebraic statistics under one geometric language, a connection the paper outlines but does not develop.
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

2 major / 4 minor

Summary. This position paper proposes the term 'neuroalgebraic geometry' for the study of neuromanifolds of algebraic machine-learning models, i.e., models whose parameterization is polynomial in parameters and inputs, so that the resulting function space is a semi-algebraic variety. The paper sets out a dictionary connecting invariants such as dimension, degree, covering number, singularities, fibers, and critical points to expressivity, sample complexity, implicit bias, identifiability, and training dynamics. It reviews a substantial body of recent work, including tensor/rank models, linear and polynomial networks, CNNs, attention mechanisms, and determinantal varieties, and illustrates the framework with an explicit two-layer linear network in Appendix A and a sample-complexity argument in Appendix B. It also contains a short section on extending the framework beyond polynomial activations via polynomial approximation and via tropical geometry for ReLU networks.

Significance. The paper's core dictionary is plausible and well supported by the cited literature, and the concrete computations in Appendices A and B are useful illustrations for a position paper. The framing has real value for the community: it collects scattered results under one proposed research program and identifies open problems. The paper does not claim to prove new theorems about the dictionary itself, which is appropriate for an invitation. However, the paper's breadth claim, namely that algebraic models can approximate arbitrary neuromanifolds and hence that algebraic tools extend to continuous activations, rests on the one-paragraph argument in Section 5.1. That argument is not rigorous and, more seriously, cannot transfer the non-metric invariants such as degree, singularities, Euclidean distance degree, and data discriminants merely by Hausdorff approximation. This is the paper's weakest load-bearing point and needs to be fixed or explicitly re-scoped.

major comments (2)
  1. [Section 5.1 (and Section 1.1)] The central transfer claim is not established. The paragraph beginning 'More precisely, consider the example...' asserts that uniform approximation of a continuous activation sigma by polynomials on a compact interval yields an approximation of the neuromanifold M_sigma by the neuromanifold of an algebraic model in Hausdorff distance. This requires a uniform continuity estimate for the map sigma -> phi_sigma over the compact parameter space, with constants that do not blow up with depth or weight bounds; no such estimate or proof is given. More importantly, even if d_H(M_sigma, M_p) -> 0, the invariants featured in the dictionary, such as degree, singularities, Euclidean distance degree, and data discriminants, are not continuous under Hausdorff convergence: a small polynomial perturbation can smooth a cusp or change the degree of the Zariski closure. Thus the abstract's promise to 'approximate arbitrary neuromanifolds' and to extend 'results and techniques' is only justified for metric quantities such as covering numbers, as in the Zhang-Kileel example cited at the end of the section. The authors should either prove the transfer with explicit hypotheses or explicitly restrict the extension claim to metric and covering-number results.
  2. [Section 1.1 (and Section 2.1)] The sentence 'algebraic models are not only general, but can approximate arbitrary neuromanifolds' conflates two different statements. The Weierstrass theorem cited in Section 5.1 approximates individual continuous functions by polynomials; it does not, by itself, approximate the image of a parameterization map, which is a set of functions indexed by parameters. Since this sentence is used in Section 2.1 to contrast neuroalgebraic geometry with kernel methods ('which can approximate arbitrary neuromanifolds'), the overstatement is load-bearing for the paper's framing. If Section 5.1 is not made rigorous, this sentence and the corresponding abstract-level promise should be weakened to say that certain metric properties of neuromanifolds of continuous activations can be bounded using algebraic approximations.
minor comments (4)
  1. [Section 2] In the first sentence of Section 2, 'relvance' should be 'relevance'.
  2. [Appendix B] The metric in Theorem B.1 is denoted by d, while Section 4.1 uses d for the degree of a variety; this notational collision makes equations such as (13) harder to read. Consider denoting the metric by rho or another symbol.
  3. [Section 4.1] Calling the degree 'an algebraic measure of how curved' the variety is imprecise; degree is an intersection-theoretic invariant and is not a curvature measure in any metric sense.
  4. [Section 5.1] The citation for the Weierstrass Approximation Theorem (de la Cerda, 2023) is an unusual source; a standard analysis textbook or a classical reference would be more helpful for the intended interdisciplinary audience.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the paper is a position/review mapping algebraic-geometric invariants to ML concepts, and its central dictionary is supported by independent prior theorems rather than by fitting or self-referential construction.

full rationale

The paper does not derive the dictionary from assumptions that already contain it. Its basic claim—that neuromanifolds of algebraic models are semi-algebraic varieties—follows from Tarski–Seidenberg (cited to Bierstone & Milman), not from an equation defined to produce the conclusion. The dimension/degree/covering-number chain (Eq. 4, Eq. 5, Eq. 13) combines known results (Kileel et al. 2019; Cucker & Smale 2002; Pontil 2003; Weyl 1939) with standard algebraic geometry; even though some of those papers share authors with this one, the cited statements are concrete theorems with proofs and are not invoked as unexamined uniqueness claims. The singularity/subnetwork statements are reported as results of prior work (Trager et al. 2020; Kohn et al. 2022; Shahverdi et al. 2025a/b) and are testable independently of the present paper's thesis. Section 5.1's transfer from polynomial to continuous activations is a legitimate external approximation theorem (Weierstrass); the text explicitly limits the transfer to 'certain cases' and points to a concrete prior application (Zhang & Kileel 2023). If anything, Section 5.1 under-proves the preservation of degree/singularities under Hausdorff approximation, but an omitted or sketched proof is a correctness/completeness concern, not a circular one. The paper contains no fitted parameter that is later renamed a prediction and no 'uniqueness theorem' imported from the authors to force a choice. Therefore no circular step is exhibited.

Assumptions & free parameters 0 free parameters · 6 assumptions · 0 invented entities

The paper relies on standard results from semi-algebraic geometry and classical theorems (Tarski-Seidenberg, fiber-dimension, Weierstrass), plus several specific bounds and structural results from the prior literature, much of it authored by the same group. No free parameters are fitted and no new entities are postulated.

assumptions (6)
  • standard math Tarski-Seidenberg theorem ensures the image of a polynomial parametrization is a semi-algebraic variety.
    Invoked in Section 3.1 to justify that neuromanifolds of algebraic models are semi-algebraic sets.
  • standard math Fiber-dimension theorem: the dimension of the image equals the co-dimension of the generic fiber.
    Used in Section 4.3 to compute neuromanifold dimension from fiber dimensions.
  • standard math Weierstrass approximation theorem: polynomials are dense in continuous functions on compact sets.
    Foundational for the approximation argument in Section 5.1 that algebraic models approximate general continuous models.
  • domain assumption Covering number bound log N_eps(M) = O(m log(d/eps) + C) for algebraic varieties.
    Equation 4 is taken from Kileel et al. 2019; the paper relies on it to connect dimension and degree to sample complexity.
  • domain assumption The neuromanifold and relevant spaces are compact or can be restricted to compact subspaces.
    Stated in Section 5.1 and needed for the covering number bound and Theorem B.1 (compactness of M).
  • domain assumption Singularities of the neuromanifold correspond to subnetworks and induce implicit bias.
    Section 4.2 attributes this to prior empirical and theoretical results; the paper treats it as a general principle of the framework, not a theorem.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Algebra Unveils Deep Learning -- An Invitation to Neuroalgebraic Geometry." pith.science (2026). https://pith.science/paper/KSTXNQSM

@misc{pith2026250118915,
  author       = {Pith},
  title        = {Pith review of: Algebra Unveils Deep Learning -- An Invitation to Neuroalgebraic Geometry},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/KSTXNQSM}},
  note         = {Machine review of arXiv:2501.18915}
}
read the original abstract

In this position paper, we promote the study of function spaces parameterized by machine learning models through the lens of algebraic geometry. To this end, we focus on algebraic models, such as neural networks with polynomial activations, whose associated function spaces are semi-algebraic varieties. We outline a dictionary between algebro-geometric invariants of these varieties, such as dimension, degree, and singularities, and fundamental aspects of machine learning, such as sample complexity, expressivity, training dynamics, and implicit bias. Along the way, we review the literature and discuss ideas beyond the algebraic domain. This work lays the foundations of a research direction bridging algebraic geometry and deep learning, that we refer to as neuroalgebraic geometry.

Figures

Figures reproduced from arXiv: 2501.18915 by the authors.

Figure 1
Figure 1. A neural variation of a celebrated doodle from the alge￾braic geometry literature (Grothendieck, 1968). model towards an estimate of the ground-truth function. Consequently, geometric problems over neuromanifolds, such as nearest point problems, govern the training dynam￾ics and provide insights into how neural networks learn. Therefore, understanding the geometry of neuromanifolds offers a twofold potential. First,… view at source ↗
Figure 2
Figure 2. A manifold covered by balls. volume of the tubular neighborhood Mε consisting of points in the ambient space of M at a distance less than ε from a point in M: Vol(Mε) ≤ Nε(M) ω2ε, (5) where ω2ε is the volume of a ball of radius 2ε. When M is a neuromanifold, the volume of Mε measures the set of functions that can be approximated by a model within a (quadratic) error of ε, which can be seen as a form of approximate e… view at source ↗
Figure 3
Figure 3. A singularity of a surface (pink) with a 3-dimensional Voronoi cell (gray). 4.2. Singularities As mentioned in Section 1.1, algebraic varieties differ from differentiable manifolds in that they may exhibit singular￾ities. A point on a variety is called singular if its tan￾gent space has a higher dimension than that of a generic point. Singularities manifest in various forms such as self￾intersections or sharp cuspid… view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Illustration of a curve with the corresponding data dis￾criminant (dashed) and Voronoi cell of the singularity (gray). For points inside (resp. outside) the discriminant, the Euclidean dis￾tance from the curve has 2 (resp. 4) real critical points. Meroni et al., 2024) …
Figure 5
Figure 5. Figure 5: The intersection of a line (dashed) – identified by two (hyper-) planes – with a surface yields a finite number of solutions, bounded by the degree of the variety. Theorem B.1. Let d be a metric on V and suppose that: • there exists C ∈ R>0 such that for every probabil…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

68 extracted references · 52 canonical work pages

  1. [1]

    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 global.max substring 't := if while FUNCTION format.date year duplicate empty "emp...

  2. [2]

    and Heaton, A

    Alexandr, Y. and Heaton, A. Logarithmic voronoi cells. Algebraic Statistics, 12 0 (1): 0 75--95, 2021

    work page 2021

  3. [3]

    Information geometry and its applications, volume 194

    Amari, S.-i. Information geometry and its applications, volume 194. Springer, 2016

    2016

  4. [4]

    Singularities affect dynamics of learning in neuromanifolds

    Amari, S.-i., Park, H., and Ozeki, T. Singularities affect dynamics of learning in neuromanifolds. Neural computation, 18 0 (5): 0 1007--1065, 2006

    work page 2006

  5. [5]

    Geometry and optimization of shallow polynomial networks

    Arjevani, Y., Bruna, J., Kileel, J., Polak, E., and Trager, M. Geometry and optimization of shallow polynomial networks. arXiv preprint arXiv:2501.06074, 2025

    arXiv 2025

  6. [6]

    Understanding deep neural networks with rectified linear units

    Arora, R., Basu, A., Mianjy, P., and Mukherjee, A. Understanding deep neural networks with rectified linear units. In International Conference on Learning Representations, 2018 a

    work page 2018

  7. [7]

    On the optimization of deep networks: Implicit acceleration by overparameterization

    Arora, S., Cohen, N., and Hazan, E. On the optimization of deep networks: Implicit acceleration by overparameterization. In International Conference on Machine Learning, pp.\ 244--253. PMLR, 2018 b

    work page 2018

  8. [8]

    and Lerario, A

    Basu, S. and Lerario, A. Hausdorff approximations and volume of tubes of singular algebraic sets. Mathematische Annalen, 387 0 (1-2): 0 79--109, 2023

    work page 2023

Show all 68 references
  1. [9]

    Voronoi diagrams of algebraic varieties under polyhedral norms

    Becedas, A., Kohn, K., and Venturello, L. Voronoi diagrams of algebraic varieties under polyhedral norms. Journal of Symbolic Computation, 120: 0 102229, 2024

    2024

  2. [10]

    and Milman, P

    Bierstone, E. and Milman, P. D. Semianalytic and subanalytic sets. Publications Math \'e matiques de l'IH \'E S , 67: 0 5--42, 1988

    1988

  3. [11]

    Rational neural networks

    Boull \'e , N., Nakatsukasa, Y., and Townsend, A. Rational neural networks. Advances in Neural Information Processing Systems, 33: 0 14243--14253, 2020

    2020

  4. [12]

    Decomposition polyhedra of piecewise linear functions

    Brandenburg, M.-C., Grillo, M., and Hertrich, C. Decomposition polyhedra of piecewise linear functions. arXiv preprint arXiv:2410.04907, 2024 a

    2024 arXiv

  5. [13]

    The real tropical geometry of neural networks for binary classification

    Brandenburg, M.-C., Loho, G., and Montufar, G. The real tropical geometry of neural networks for binary classification. Transactions on Machine Learning Research, 2024 b

    2024

  6. [14]

    Metric Algebraic Geometry

    Breiding, P., Kohn, K., and Sturmfels, B. Metric Algebraic Geometry. Oberwolfach Seminars. Birkhäuser, Basel, 2024

    2024

  7. [15]

    M., Bruna, J., Cohen, T., and Veli c kovi \'c , P

    Bronstein, M. M., Bruna, J., Cohen, T., and Veli c kovi \'c , P. Geometric deep learning: Grids, groups, graphs, geodesics, and gauges. arXiv preprint arXiv:2104.13478, 2021

    2021 arXiv

  8. [16]

    and Vetter, U

    Bruns, W. and Vetter, U. Determinantal rings, volume 1327. Springer, 2006

    2006

  9. [17]

    Neuromanifolds

    Calin, O. Neuromanifolds. Deep Learning Architectures: A Mathematical Approach, pp.\ 465--504, 2020

    2020

  10. [18]

    The maximum likelihood degree

    Catanese, F., Ho s ten, S., Khetan, A., and Sturmfels, B. The maximum likelihood degree. American Journal of Mathematics, 128 0 (3): 0 671--697, 2006

    2006

  11. [19]

    \"O ., Jamneshan, A., Mont \'u far, G., Sturmfels, B., and Venturello, L

    C elik, T. \"O ., Jamneshan, A., Mont \'u far, G., Sturmfels, B., and Venturello, L. Optimal transport to a variety. In Mathematical Aspects of Computer and Information Sciences (MACIS 2019), Revised Selected Papers, pp.\ 364--381. Springer, 2020

    2019

  12. [20]

    and Smale, S

    Cucker, F. and Smale, S. On the mathematical foundations of learning. Bulletin of the American Mathematical Society, 39 0 (1): 0 1--49, 2002

    2002

  13. [21]

    Polynomial approximations to continuous functions

    de la Cerda, S. Polynomial approximations to continuous functions. The American Mathematical Monthly, 130 0 (7): 0 655--655, 2023

    2023

  14. [22]

    Draisma, J., Horobe t , E., Ottaviani, G., Sturmfels, B., and Thomas, R. R. The euclidean distance degree of an algebraic variety. Foundations of Computational Mathematics, 16: 0 99--149, 2016

    2016

  15. [23]

    S., Hu, W., and Lee, J

    Du, S. S., Hu, W., and Lee, J. D. Algorithmic regularization in learning deep homogeneous models: Layers are automatically balanced. Advances in Neural Information Processing Systems, 31, 2018

    2018

  16. [24]

    and Harer, J

    Edelsbrunner, H. and Harer, J. Computational Topology: an Introduction. American Mathematical Soc., 2010

    2010

  17. [25]

    and Harris, J

    Eisenbud, D. and Harris, J. On varieties of minimal degree. In Proc. Sympos. Pure Math, volume 46, pp.\ 3--13, 1987

    1987

  18. [26]

    Fefferman, C. et al. Reconstructing a neural net from its output. Revista Matem \'a tica Iberoamericana , 10 0 (3): 0 507--556, 1994

    1994

  19. [27]

    I., Wu, C., and Yahl, T

    Finkel, B., Rodriguez, J. I., Wu, C., and Yahl, T. Activation thresholds and expressiveness of polynomial neural networks. arXiv preprint arXiv:2408.04569, 2024

    2024

  20. [28]

    and Carbin, M

    Frankle, J. and Carbin, M. The lottery ticket hypothesis: Finding sparse, trainable neural networks. In International Conference on Learning Representations, 2019

    2019

  21. [29]

    Intersection theory, volume 2

    Fulton, W. Intersection theory, volume 2. Springer Science & Business Media, 2013

    2013

  22. [30]

    G., and Veli c kovi \'c , P

    Gavranovi \'c , B., Lessard, P., Dudzik, A., von Glehn, T., Ara \'u jo, J. G., and Veli c kovi \'c , P. Categorical deep learning: An algebraic theory of architectures. arXiv preprint arXiv:2402.15332, 2024

    2024 arXiv

  23. [31]

    Hidden symmetries of relu networks

    Grigsby, E., Lindsey, K., and Rolnick, D. Hidden symmetries of relu networks. In International Conference on Machine Learning, pp.\ 11734--11760. PMLR, 2023

    2023

  24. [32]

    Classes de faisceaux et theoreme de rimann-roch

    Grothendieck, A. Classes de faisceaux et theoreme de rimann-roch. 1968

    1968

  25. [33]

    N., Birdal, T., Dey, T

    Hajij, M., Zamzmi, G., Papamarkou, T., Miolane, N., Guzm \'a n-S \'a enz, A., Ramamurthy, K. N., Birdal, T., Dey, T. K., Mukherjee, S., Samaga, S. N., et al. Topological deep learning: Going beyond graph data. arXiv preprint arXiv:2206.00606, 2022

    2022 arXiv

  26. [34]

    W., Marchetti, G

    Henry, N. W., Marchetti, G. L., and Kohn, K. Geometry of lightning self-attention: Identifiability and dimension. In International Conference on Learning Representations, 2025

    2025

  27. [35]

    A., Jordan, M

    Jacobs, R. A., Jordan, M. I., Nowlan, S. J., and Hinton, G. E. Adaptive mixtures of local experts. Neural computation, 3 0 (1): 0 79--87, 1991

    1991

  28. [36]

    Neural tangent kernel: Convergence and generalization in neural networks

    Jacot, A., Gabriel, F., and Hongler, C. Neural tangent kernel: Convergence and generalization in neural networks. Advances in Neural Information Processing Systems, 31, 2018

    2018

  29. [37]

    On the expressive power of deep polynomial neural networks

    Kileel, J., Trager, M., and Bruna, J. On the expressive power of deep polynomial neural networks. Advances in Neural Information Processing Systems, 32, 2019

    2019

  30. [38]

    The geometry of the neuromanifold

    Kohn, K. The geometry of the neuromanifold. SIAM News, 57 0 (06), 2024

    2024

  31. [39]

    Geometry of linear convolutional networks

    Kohn, K., Merkh, T., Mont \'u far, G., and Trager, M. Geometry of linear convolutional networks. SIAM Journal on Applied Algebra and Geometry, 6 0 (3): 0 368--406, 2022

    2022

  32. [40]

    Function space and critical points of linear convolutional networks

    Kohn, K., Mont \'u far, G., Shahverdi, V., and Trager, M. Function space and critical points of linear convolutional networks. SIAM Journal on Applied Algebra and Geometry, 8 0 (2): 0 333--362, 2024

    2024

  33. [41]

    Geometry of polynomial neural networks

    Kubjas, K., Li, J., and Wiesmann, M. Geometry of polynomial neural networks. Algebraic Statistics, 15 0 (2): 0 295--328, 2024

    2024

  34. [42]

    Lasserre, J. B. Global optimization with polynomials and the problem of moments. SIAM Journal on optimization, 11 0 (3): 0 796--817, 2001

    2001

  35. [43]

    and Sturmfels, B

    Maclagan, D. and Sturmfels, B. Introduction to tropical geometry, volume 161. American Mathematical Society, 2021

    2021

  36. [44]

    L., Hillar, C

    Marchetti, G. L., Hillar, C. J., Kragic, D., and Sanborn, S. Harmonics of learning: Universal fourier features emerge in invariant networks. In The Thirty Seventh Annual Conference on Learning Theory, pp.\ 3775--3797. PMLR, 2024

    2024

  37. [45]

    Abide by the law and follow the flow: Conservation laws for gradient flows

    Marcotte, S., Gribonval, R., and Peyr \'e , G. Abide by the law and follow the flow: Conservation laws for gradient flows. Advances in Neural Information Processing Systems, 36, 2024

    2024

  38. [46]

    Low rank approximation: algorithms, implementation, applications, volume 906

    Markovsky, I. Low rank approximation: algorithms, implementation, applications, volume 906. Springer, 2012

    2012

  39. [47]

    The algebraic degree of the wasserstein distance

    Meroni, C., Reinke, B., and Wang, K. The algebraic degree of the wasserstein distance. arXiv preprint arXiv:2401.12735, 2024

    2024 arXiv

  40. [48]

    Milnor, J. W. Morse theory. Number 51. Princeton university press, 1963

    1963

  41. [49]

    Sharp bounds for the number of regions of maxout networks and vertices of minkowski sums

    Mont \'u far, G., Ren, Y., and Zhang, L. Sharp bounds for the number of regions of maxout networks and vertices of minkowski sums. SIAM Journal on Applied Algebra and Geometry, 6 0 (4): 0 618--649, 2022

    2022

  42. [50]

    Orhan, A. E. and Pitkow, X. Skip connections eliminate singularities. In International Conference on Learning Representations, 2018

    2018

  43. [51]

    Approximation theory of the mlp model in neural networks

    Pinkus, A. Approximation theory of the mlp model in neural networks. Acta numerica, 8: 0 143--195, 1999

    1999

  44. [52]

    Pistone, G., Riccomagno, E., and Wynn, H. P. Algebraic statistics: Computational commutative algebra in statistics. Chapman and Hall/CRC, 2000

    2000

  45. [53]

    A note on different covering numbers in learning theory

    Pontil, M. A note on different covering numbers in learning theory. Journal of Complexity, 19 0 (5): 0 665--671, 2003

    2003

  46. [54]

    and Hirsch, K

    Shafarevich, I. and Hirsch, K. Basic algebraic geometry, volume 1. Springer, 1994

    1994

  47. [55]

    Algebraic complexity and neurovariety of linear convolutional networks

    Shahverdi, V. Algebraic complexity and neurovariety of linear convolutional networks. arXiv preprint arXiv:2401.16613, 2024

    2024 arXiv

  48. [56]

    L., and Kohn, K

    Shahverdi, V., Marchetti, G. L., and Kohn, K. On the geometry and optimization of polynomial convolutional networks. In Artificial Intelligence and Statistics, 2025 a

    2025

  49. [57]

    L., and Kohn, K

    Shahverdi, V., Marchetti, G. L., and Kohn, K. Learning on a razor's edge: the singularity bias of polynomial neural networks. arXiv preprint arXiv:2505.11846, 2025 b

    2025 arXiv

  50. [58]

    Pure and spurious critical points: a geometric study of linear networks

    Trager, M., Kohn, K., and Bruna, J. Pure and spurious critical points: a geometric study of linear networks. In International Conference on Learning Representations, 2020

    2020

  51. [59]

    Tran, N. M. and Wang, J. Minimal representations of tropical rational functions. Algebraic Statistics, 15 0 (1): 0 27--59, 2024

    2024

  52. [60]

    Attention is all you need

    Vaswani, A. Attention is all you need. Advances in Neural Information Processing Systems, 2017

    2017

  53. [61]

    Algebraic geometry and statistical learning theory, volume 25

    Watanabe, S. Algebraic geometry and statistical learning theory, volume 25. Cambridge university press, 2009

    2009

  54. [62]

    On the volume of tubes

    Weyl, H. On the volume of tubes. American Journal of Mathematics, 61 0 (2): 0 461--472, 1939

    1939

  55. [63]

    Gradient dynamics of shallow univariate relu networks

    Williams, F., Trager, M., Panozzo, D., Silva, C., Zorin, D., and Bruna, J. Gradient dynamics of shallow univariate relu networks. Advances in Neural Information Processing Systems, 32, 2019

    2019

  56. [64]

    and Comte, G

    Yomdin, Y. and Comte, G. Tame geometry with application in smooth analysis. Springer, 2004

    2004

  57. [65]

    E., and Townsend, A

    Yu, A., Becquey, C., Halikias, D., Mallory, M. E., and Townsend, A. Arbitrary-depth universal approximation theorems for operator neural networks. arXiv preprint arXiv:2109.11354, 2021

    2021 arXiv

  58. [66]

    Tropical geometry of deep neural networks

    Zhang, L., Naitzat, G., and Lim, L.-H. Tropical geometry of deep neural networks. In International Conference on Machine Learning, pp.\ 5824--5832. PMLR, 2018

    2018

  59. [67]

    and Kileel, J

    Zhang, Y. and Kileel, J. Covering number of real algebraic varieties and beyond: Improved bounds and applications. arXiv e-prints, pp.\ arXiv--2311, 2023

    2023

  60. [68]

    A survey on model compression for large language models

    Zhu, X., Li, J., Liu, Y., Ma, C., and Wang, W. A survey on model compression for large language models. Transactions of the Association for Computational Linguistics, 12: 0 1556--1577, 2024

    2024

Pith tools

Reviewed August 9, 2026 · model on record in the stance chip above.