Pith. sign in

REVIEW 5 major objections 5 minor 51 references

Signature Maximum Mean Discrepancy Two-Sample Statistical Tests

T0 review · 5 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 Path scaling turns a failing signature two-sample test into a reliable one.

desk verdict Useful level-contribution analysis and a real practical insight, but the headline power gains rely on post hoc scaling choices and the hypothesis-testing section has fixable but real errors. read the letter →

arxiv 2506.01718 v1 pith:PWPJXMT5 submitted 2025-06-02 stat.ML cs.LGmath.DS

Andrew Alden , Blanka Horvath , Zacharia Issa This is my paper
classification stat.MLcs.LGmath.DS MSC 62G1062M10
keywords signaturekernelmaximummeandiscrepancytwo-sampletestingstochasticprocessespathType2errorembeddingGARCH
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 tries to establish that the signature maximum mean discrepancy (sig-MMD) can serve as a practical two-sample test for stochastic processes, and that its main failure mode—accepting different processes as the same—is largely fixable. The fix is to scale the input paths before computing the signature kernel, because scaling is equivalent to reweighting the higher-order signature levels where distributional differences actually show up. On scaled Brownian motions with different volatilities, path scaling by 3 drops the Type 2 error from 72.6% to 2.0%; on GARCH models, scaling by 5.5 drops it from 90.2% to 0.0%, while Type 1 error stays near the 5% significance level. This matters because path-valued two-sample tests are useful in finance and time-series modelling, where limited data makes false agreement costly.

What carries the argument

The load-bearing identity is the scalar-multiplication equivalence for signature kernels: $k_{\mathrm{Sig}}(\theta x, y) = k_{\mathrm{Sig}}(x, \theta y) = k^{\varphi}_{\mathrm{Sig}}(x, y)$ with $\varphi(m)=\theta^m$, which makes path scaling identical to weighting level-$m$ signature inner products by $\theta^m$. The paper also decomposes the $\varphi$-MMD into per-level contributions $\Gamma^{\varphi}_m$, so one can see which signature level carries the separation between the null and alternative distributions; the factorial decay estimate $\|\Phi_{\mathrm{Sig},m}(x)\| \le \|x\|_p^m / m!$ explains why, without reweighting, lower levels that do not separate the distributions dominate the statistic.

What would settle it

Run the Section 4.3 mixture-model pair—two processes with nearly equal first and second moments but different third moment—at batch size 128 and compute the per-level contributions under increasing scaling; if no finite scaling makes the third-level term separate from noise, then scaling reweights levels but cannot by itself make high-moment differences detectable in low-data regimes, which the paper's own 94.6% Type 2 error already hints at.

Watch

Extended reading notes

Core claim

The central claim is that the default sig-MMD two-sample test is systematically underpowered when two stochastic processes differ mainly in higher-order moments, because signature terms decay factorially and low-level terms dominate the statistic. Scaling both collections of paths by a common factor $\sqrt{\theta}$ is not a neutral preprocessing step: by [9, Corollary 2.10] it is exactly the $\varphi$-signature kernel with weight $\varphi(m)=\theta^m$ on level-$m$ signature terms, so it reweights the statistic toward the levels that separate the distributions. With this reweighting, the test detects differences between scaled Brownian motions and between GARCH models at batch size 128 with Type 2 errors near zero and Type 1 errors near the nominal 5%, and it detects distributional differences at much smaller batch sizes than the unscaled test.

Load-bearing premise

The power gains assume the user can find a good scaling factor, RBF bandwidth, and preprocessing pipeline without knowing the true relationship between the samples; in the paper every successful scaling (3, 5.5, 2, 0.8, 5) was selected by inspecting Type 2 error curves, and the Section 5 calibration estimates Type 2 error under an assumed alternative on unlabelled data.

Editorial extensions

If this is right

  • For scaled Brownian motions with volatilities 0.2 and 0.3 and batch size 128, scaling by 3 lowers Type 2 error from 72.6% to 2.0% with the biased estimator and from 85.8% to 6.6% with the unbiased estimator.
  • For GARCH models with distinct parameters, scaling by 5.5 lowers Type 2 error from 90.2% to 0.0% at batch size 128.
  • Type 1 error remains centered near the 5% significance level across scaling factors and batch sizes, for both biased and unbiased estimators.
  • Scaling lets the test separate distributions at much smaller batch sizes, which matters when data are scarce or computation is expensive.
  • The per-level decomposition shows which signature level is responsible for separation, guiding the choice of scaling and preprocessing before the test is run.

Reading between the lines

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

  • Because scaling is equivalent to changing the weight of each signature level, the same data expressed in different units can lead to different test conclusions; practitioners should treat scaling as a tunable hyperparameter, not an arbitrary normalization.
  • The reweighting idea is not tied to the signature kernel: any graded feature map whose terms decay across levels could be reweighted analogously to emphasize the levels where two distributions differ.
  • When paths are lifted through an RBF kernel, the scaling must be applied to the lifted path rather than the raw path, or the level-reweighting effect is lost.
  • The mixture-model example suggests a diagnostic use of the level decomposition: checking whether any finite scaling makes the separating level dominate tells you in advance whether the test can succeed at a given batch size.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

5 major / 5 minor

Summary. The paper studies the signature Maximum Mean Discrepancy (sig-MMD) as a test statistic for two-sample tests on path-valued data. It introduces the phi-weighted signature kernel and the corresponding phi-MMD, decomposes the MMD into signature-level contributions, and argues that path scaling reweights higher-order signature levels, thereby reducing Type 2 errors. Through simulations with scaled Brownian motions, GARCH processes, and mixture models, the authors report large power improvements from scaling (e.g., Type 2 error falling from 72.6% to 2.0% for Brownian motion and from 90.2% to 0.0% for GARCH), while Type 1 error remains near the 5% significance level. The final section applies the pipeline to sector return data and concludes that the distributions are different. Public code is provided.

Significance. If the reported power gains were reproducible under a fixed protocol, the paper would be practically useful: it gives accessible numerical evidence that the signature kernel's level weighting strongly affects test power, and it ships public code and transparent level-decomposition diagnostics. The checks showing Type 1 error stability across scaling factors are a useful robustness observation. However, the central empirical claim is not yet established because the scaling factors are selected post hoc on the same data used to evaluate power, and the Section 5 calibration uses labels that the test is supposed to infer. The technical errors in Section 3 also need correction before the results can be relied upon.

major comments (5)
  1. [Section 3, first paragraph after fixing distributions] The rejection rule is stated backwards: the null hypothesis is 'rejected with significance alpha if d-hat^2 < c-hat_{1-alpha}'. Since the MMD is nonnegative and large values indicate disagreement between distributions, the correct rule is rejection when d-hat^2 > c-hat_{1-alpha}. As written, the procedure rejects H0 when the statistic is small, which is inconsistent with all power calculations in Section 4 and with the Type 2 error formula in Eq. (7). This must be corrected, and the numerical code should be checked against the corrected rule.
  2. [Section 3.3, asymptotic null distribution] The asymptotic null distribution is mis-normalized: the paper states that (1/N) d-hat^2_{kSig,2} converges to sum_i lambda_i (z_i^2 - 2), but the standard result in the cited references is that N times the unbiased MMD statistic converges to that weighted chi-square sum. With the displayed 1/N factor, the right-hand side would not be a valid asymptotic threshold. Please correct the normalization and confirm that no numerical result in Section 4 relies on this formula.
  3. [Sections 4.1-4.3, scaling factor selection] The headline power improvements are obtained with scaling factors (3, 5.5, 2, 0.8, 5) that were selected by inspecting Type 2 error curves on the same simulated data used to report the power. The paper states that 'various scaling factors' were tested and then reports the value that gives the dramatic improvement; this is oracle selection. A Type 2 error rate is an operating characteristic of a fixed test procedure, so the reductions from 72.6% to 2.0% and from 90.2% to 0.0% are not yet shown to be achievable under a legitimate protocol. Please either prespecify a selection rule (e.g., a moment-based heuristic, a separate validation set, or an aggregated test over a grid) and evaluate power after selection, or explicitly reframe the reported numbers as illustrative best-case effects of scaling.
  4. [Section 5, uncontrolled environment calibration] The calibration procedure computes the probability of a Type 2 error on the 80% calibration split, but in an uncontrolled environment the true hypothesis is unknown; computing a Type 2 error requires assuming an alternative, which uses exactly the labels the test is supposed to infer. The final permutation test on the 20% holdout is a single decision with no repeated-selection analysis and no Type 1 error estimate. Consequently, the real-data conclusion that the two sector distributions are not equal is not supported as a demonstration of a valid testing procedure. Please provide a protocol that does not use alternative labels for calibration, or clarify that Section 5 is an illustration whose statistical validity is not claimed.
  5. [Section 2.5, definition of Gamma^phi_m] The displayed definition of the level-m contribution reads Gamma^phi_m(PX,PY) = phi(m)[Lambda_m(PX,PY) - 2 Lambda_m(PX,PY) + Lambda_m(PX,PY)], with identical arguments in all three terms. It should be phi(m)[Lambda_m(PX,PX) - 2 Lambda_m(PX,PY) + Lambda_m(PY,PY)] to match the MMD expansion in Eq. (6) and the estimators used in the numerical sections. As written, the object is identically zero and cannot underlie the level-contribution plots in Figs. 2, 4, 8, 10, 13, and 14.
minor comments (5)
  1. [Section 4.1, paragraph after Fig. 1] The text refers to 'the high probability 85.5%', but the preceding sentence reports 85.8% for the unbiased estimator; please make the numbers consistent.
  2. [Section 3, Eq. (7) and surrounding text] The notation dF_H0 and dF_H1 is ambiguous: if F denotes a cumulative distribution function, write F(c-hat); if dF denotes a density, Eq. (7) should be an integral rather than a point evaluation. Please clarify the notation.
  3. [Section 2.5, moment interpretation] The statement that the level-K term of the expected signature is 'associated with the K-th moment' is imprecise: signature level K contains iterated integrals such as integral integral dX^i dX^j, which mix moments of different orders and include quadratic variation terms. Since this heuristic is used to justify the scaling mechanism, please replace it with a precise statement in terms of expected iterated integrals and the fact that scaling a path by theta multiplies the level-m signature term by theta^m.
  4. [Section 5, Figs. 18 and 19] The number of bootstrap or permutation replications used in the calibration and final decision is not stated; please report these values for reproducibility.
  5. [Section 2.5, Definition 2.1] The set XSeq is used before it is defined; please define it at first use.

Circularity Check

3 steps flagged · score 6.0 of 10

Headline power gains rest on scaling factors chosen by looking at the Type 2 error on the same data; without a prespecified selection rule, the reported reductions are in-sample optima.

  1. fitted input called prediction [Section 4.1, Scaled Brownian Motion (Figs. 3 and 5)]
    "In our simulations, we tested various scaling factors. ... When re-running all simulations with scaled paths using a scaling factor of 3, the probability of a Type 2 error occurring drops to 2.0% (Fig. 3a) when using the biased estimator and 6.6% (Fig. 3b) when using the unbiased estimator."

    The Type 2 error probability is both the selection objective and the reported outcome. The paper states it 'tested various scaling factors' and then reports the value (3) that minimizes the same Type 2-error-versus-scaling curve (Fig. 5) computed on the same simulated data. The drop from 72.6% to 2.0% is therefore the minimum of the selection criterion, not the operating characteristic of a prespecified sig-MMD test. No prespecified selection rule (moment heuristic, separate validation data, or alternative-free diagnostic) is supplied, so the headline reduction is an in-sample fit.

  2. fitted input called prediction [Section 4.2, Autoregressive Time Series Models / GARCH (Figs. 7 and 9)]
    "Once again, path scaling was sufficient to reduce this probability. When a scaling of 5.5 was applied, the probability of a Type 2 error occurring dropped to 0.0% (Fig. 7b)."

    The scaling factor 5.5 is selected from the Type 2 error surface in Fig. 9, which is computed on the same GARCH simulations that later yield the reported 0.0% error. Reporting the Type 2 error at the selected scaling is reporting the best point of the fitted curve; it is not a prediction for a fixed, prespecified test procedure. The paper does not state how a user would choose 5.5 without seeing the Type 2 error curve, so the reported improvement is statistically forced by the selection.

1 more flagged steps
  1. fitted input called prediction [Section 5, Uncontrolled Environment (calibration and holdout)]
    "We propose optimising the scaling factor in a similar way, mimicking the optimisation procedure for certain hyperparameters in a ML context. Since we are focused on two-sample testing, the hyperparameters are optimised for a low probability of a Type 2 error occurring. ... 80% of the data was used to calibrate the hyperparameters of the test. ..."

    The calibration stage computes Type 2 error under an assumed alternative on the 80% labelled calibration portion, so the labels the test is supposed to infer are used to select the scaling and RBF parameter. The final 20% holdout is a single permutation-test realization; no repeated-selection evaluation or holdout Type 1/Type 2 error estimate is reported. The chosen configuration is therefore fitted to labelled calibration data, and the holdout 'conclusion' is not an independent validation of the tuning protocol.

full rationale

The mathematical core of the paper—the reformulation of sig-MMD as a sum of level contributions and the equivalence between path scaling and the weight function phi(m)=theta^m—is imported from Cass-Lyons-Xu (refs [9, 10]) and from the MMD theory of Gretton et al.; those are external results, not self-citations by the present authors, and the paper does not derive its headline effect from a self-citation chain. The central circularity is empirical. In Sections 4.1, 4.2, and 4.3, the scaling factor is selected by inspecting the Type 2 error as a function of scaling on the same simulated data (Figs. 5, 9, 16), and the reported reductions (72.6% to 2.0%; 90.2% to 0.0%; 94.6% to 21.0%) are the values of that same objective at the selected points. A Type 2 error rate is an operating characteristic of a fixed test procedure; if the procedure includes 'inspect the Type 2 error curve and pick the best theta,' then the reported error is a fitted minimum, not a prediction. Section 5 does not cure this: the calibration stage computes Type 2 error under an assumed alternative on the 80% labelled calibration split, and the final 20% holdout is a single realization with no repeated-selection evaluation. The Type 1 error checks and the level-contribution analysis have independent content, which is why the score is 6 rather than 8; but the headline power claim, as stated, reduces by construction to the selection criterion.

Assumptions & free parameters 2 free parameters · 4 assumptions · 0 invented entities

The mathematical objects are inherited from prior literature. The free parameters that drive the empirical claims are the scaling factors and RBF bandwidths, selected per example. No new physical or mathematical entities are postulated.

free parameters (2)
  • Path scaling factor theta = 3 (scaled Brownian), 5.5 (GARCH), 2 (mixture, lifted), 0.8 (lifted RBF), 5 (real data)
    Chosen by grid search and inspection of Type 2 error curves in each example; the paper claims scaling is a key lever for test power but provides no automatic selection rule.
  • RBF smoothing parameter sigma_RBF = sqrt(0.1), sqrt(0.5), 1, 0.5 depending on experiment
    Set by hand for feature-space lifted paths; changing this value changes Type 2 error as shown in Figure 16.
assumptions (4)
  • domain assumption The signature kernel is characteristic and universal on compact subsets of path space, so sig-MMD is a metric.
    Invoked in Section 2.1; relies on Chevyrev-Oberhauser and Salvi et al. results.
  • domain assumption Linearly interpolated sample paths belong to X^{1-var}, so signature theory applies to discretely sampled paths.
    Stated in Section 2.1; needed for applying sig-MMD to Brownian-like paths observed at discrete times.
  • standard math U-statistic asymptotic results for MMD, due to Gretton et al., transfer to the sig-MMD with the stated kernel h.
    Used in Sections 3.3 and 3.4 to justify the null and alternative distributions of the test statistic.
  • standard math Donsker's invariance principle justifies treating linearly interpolated Brownian paths as converging to Brownian motion.
    Mentioned at the start of Section 4 to connect simulations on discrete paths to continuous processes.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Signature Maximum Mean Discrepancy Two-Sample Statistical Tests." pith.science (2026). https://pith.science/paper/PWPJXMT5

@misc{pith2026250601718,
  author       = {Pith},
  title        = {Pith review of: Signature Maximum Mean Discrepancy Two-Sample Statistical Tests},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/PWPJXMT5}},
  note         = {Machine review of arXiv:2506.01718}
}
read the original abstract

Maximum Mean Discrepancy (MMD) is a widely used concept in machine learning research which has gained popularity in recent years as a highly effective tool for comparing (finite-dimensional) distributions. Since it is designed as a kernel-based method, the MMD can be extended to path space valued distributions using the signature kernel. The resulting signature MMD (sig-MMD) can be used to define a metric between distributions on path space. Similarly to the original use case of the MMD as a test statistic within a two-sample testing framework, the sig-MMD can be applied to determine if two sets of paths are drawn from the same stochastic process. This work is dedicated to understanding the possibilities and challenges associated with applying the sig-MMD as a statistical tool in practice. We introduce and explain the sig-MMD, and provide easily accessible and verifiable examples for its practical use. We present examples that can lead to Type 2 errors in the hypothesis test, falsely indicating that samples have been drawn from the same underlying process (which generally occurs in a limited data setting). We then present techniques to mitigate the occurrence of this type of error.

Figures

Figures reproduced from arXiv: 2506.01718 by the authors.

Figure 1
Figure 1. Null and alternative distributions of the (squared) sig-MMD between two scaled [PITH_FULL_IMAGE:figures/full_fig_p021_1.png] view at source ↗
Figure 2
Figure 2. Level contributions to the sig-MMD between two scaled Brownian motions. Batch [PITH_FULL_IMAGE:figures/full_fig_p021_2.png] view at source ↗
Figure 3
Figure 3. Null and alternative distributions of the (squared) [PITH_FULL_IMAGE:figures/full_fig_p022_3.png] view at source ↗
Figures from the paper (16 more)
Figure 4
Figure 4. Figure 4: Level contributions to the 𝜙-MMD between two scaled Brownian motions. Batch size of 128 was used and 2,048 independent simulations were run. A scaling of 3 was applied and the unbiased estimator was used. In the above example, the batch size was kept fixed at 128. It i…
Figure 5
Figure 5. Figure 5: Probability of a Type 2 error occurring as a function of scaling factor and batch [PITH_FULL_IMAGE:figures/full_fig_p024_5.png]
Figure 6
Figure 6. Figure 6: Probability of a Type 1 error occurring as a function of sample size and scaling [PITH_FULL_IMAGE:figures/full_fig_p025_6.png]
Figure 7
Figure 7. Figure 7: Null and alternative distributions of the [PITH_FULL_IMAGE:figures/full_fig_p028_7.png]
Figure 8
Figure 8. Figure 8: Null and alternative distributions of the level contributions between two GARCH [PITH_FULL_IMAGE:figures/full_fig_p028_8.png]
Figure 9
Figure 9. Figure 9: Probability of a Type 2 error occurring between two GARCH models as a function [PITH_FULL_IMAGE:figures/full_fig_p029_9.png]
Figure 10
Figure 10. Figure 10: Distribution of the level terms between two GARCH models as a function of [PITH_FULL_IMAGE:figures/full_fig_p030_10.png]
Figure 11
Figure 11. Figure 11: Probability of a Type 1 error occurring between two GARCH models as a function [PITH_FULL_IMAGE:figures/full_fig_p031_11.png]
Figure 12
Figure 12. Figure 12: Null and alternative distributions of the [PITH_FULL_IMAGE:figures/full_fig_p032_12.png]
Figure 13
Figure 13. Figure 13: Null and alternative distributions of the level contributions between two mixture [PITH_FULL_IMAGE:figures/full_fig_p033_13.png]
Figure 14
Figure 14. Figure 14: Null and alternative distributions of the level contributions between two mixture [PITH_FULL_IMAGE:figures/full_fig_p034_14.png]
Figure 15
Figure 15. Figure 15: Null and alternative distributions of the [PITH_FULL_IMAGE:figures/full_fig_p035_15.png]
Figure 16
Figure 16. Figure 16: Probability of a Type 2 error occurring between two mixture models as a function [PITH_FULL_IMAGE:figures/full_fig_p036_16.png]
Figure 17
Figure 17. Figure 17: Probability of a Type 1 error occurring between two mixture models as a function [PITH_FULL_IMAGE:figures/full_fig_p037_17.png]
Figure 18
Figure 18. Figure 18: Null and alternative distributions of the [PITH_FULL_IMAGE:figures/full_fig_p039_18.png]
Figure 19
Figure 19. Figure 19: Null and alternative distributions of the [PITH_FULL_IMAGE:figures/full_fig_p040_19.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

51 extracted references · 51 canonical work pages

  1. [1]

    King’s Computational Research, Engineering and Technology Environment (CREATE),

    work page

  2. [2]

    J. Ai, O. Kuˇ zelka, and Y. Wang. Hoeffding–Serfling Inequality for U-Statistics Without Replacement. Journal of Theoretical Probability, 36:390–408, 2023

    work page 2023

  3. [3]

    Alden, C

    A. Alden, C. Ventre, B. Horvath, and G. Lee. Model-Agnostic Pricing of Exotic Deriva- tives Using Signatures. In Proceedings of the Third ACM International Conference on AI in Finance , ICAIF ’22, page 96–104, New York, NY, USA, 2022. Association for Computing Machinery

    work page 2022

  4. [4]

    Andr` es, A

    H. Andr` es, A. Boumezoued, and B. Jourdain. Signature-based validation of real-world economic scenarios. ASTIN Bulletin , 54(2):410–440, 2024

    work page 2024

  5. [5]

    Beutner and H

    E. Beutner and H. Z¨ ahle. Deriving the asymptotic distribution of U- and V-statistics of dependent data using weighted empirical processes. Bernoulli, 18(3):803 – 822, 2012

    work page 2012

  6. [6]

    Black and M

    F. Black and M. Scholes. The Pricing of Options and Corporate Liabilities. Journal of Political Economy, 81(3):637–654, 1973

    work page 1973

  7. [7]

    Bollerslev

    T. Bollerslev. Generalized autoregressive conditional heteroskedasticity. Journal of Econometrics, 31(3):307–327, 1986

    work page 1986

  8. [8]

    B¨ uhler, B

    H. B¨ uhler, B. Horvath, T. Lyons, I. P. Arribas, and B. Wood. A Data-Driven Market Simulator for Small Data Environments. 2020

    work page 2020

Show all 51 references
  1. [9]

    T. Cass, T. Lyons, and X. Xu. General Signature Kernels. 2021

    2021

  2. [10]

    T. Cass, T. Lyons, and X. Xu. Weighted signature kernels. The Annals of Applied Probability, 34(1A):585 – 626, 2024

    2024

  3. [11]

    Ch´ erief-Abdellatif and P

    B.-E. Ch´ erief-Abdellatif and P. Alquier. MMD-Bayes: Robust Bayesian Estimation via Maximum Mean Discrepancy. In Proceedings of The 2nd Symposium on Advances in Approximate Bayesian Inference , volume 118 of Proceedings of Machine Learning Research (PMLR), pages 1–21. PMLR, 0...

    2020

  4. [12]

    Chevyrev and A

    I. Chevyrev and A. Kormilitzin. A Primer on the Signature Method in Machine Learn- ing. 2016

    2016

  5. [13]

    Chevyrev and H

    I. Chevyrev and H. Oberhauser. Signature Moments to Characterize Laws of Stochastic Processes. Journal of Machine Learning Research , 23(176):1–42, 2022

    2022

  6. [14]

    M. D. Donsker. An Invariance Principle for Certain Probability Limit Theorems. Amer- ican Mathematical Society. Memoirs. 1951

    1951

  7. [15]

    R. F. Engle. Autoregressive Conditional Heteroscedasticity with Estimates of the Vari- ance of United Kingdom Inflation. Econometrica, 50(4):987–1007, 1982

    1982

  8. [16]

    R. F. Engle and A. J. Patton. What good is a volatility model? Quantitative Finance, 1(2):237–245, 2001

    2001

  9. [17]

    T. S. Ferguson. U-statistics notes for statistics 200c, 2005

    2005

  10. [18]

    Fermanian

    A. Fermanian. Embedding and learning with signatures. Computational Statistics & Data Analysis, 157, 2021

    2021

  11. [19]

    X. Geng, H. Ni, and C. Wang. Expected Signature on a Riemannian Manifold and Its Geometric Implications. 2024

    2024

  12. [20]

    Gretton, K

    A. Gretton, K. Borgwardt, M. Rasch, B. Sch¨ olkopf, and A. Smola. A Kernel Method for the Two-Sample-Problem. In Advances in Neural Information Processing Systems , volume 19. MIT Press, 2006

    2006

  13. [21]

    Gretton, K

    A. Gretton, K. Borgwardt, M. Rasch, B. Sch¨ olkopf, and A. Smola. A Kernel Two- Sample Test. Journal Of Machine Learning Research , 13(25):723–773, 2012

    2012

  14. [22]

    Gretton, K

    A. Gretton, K. Fukumizu, Z. Harchaoui, and B. K. Sriperumbudur. A Fast, Consis- tent Kernel Two-Sample Test. In Advances in Neural Information Processing Systems , volume 22. Curran Associates, Inc., 2009

    2009

  15. [23]

    Gretton, D

    A. Gretton, D. Sejdinovic, H. Strathmann, S. Balakrishnan, M. Pontil, K. Fukumizu, and B. K. Sriperumbudur. Optimal kernel choice for large-scale two-sample tests. In 42 Advances in Neural Information Processing Systems , volume 25. Curran Associates, Inc., 2012

    2012

  16. [24]

    L. G. Gyurk´ o, T. Lyons, M. Kontkowski, and J. Field. Extracting information from the signature of a financial data stream. 2014

    2014

  17. [25]

    Harrison, D

    D. Harrison, D. Sutton, P. Carvalho, and M. Hobson. Validation of Bayesian posterior distributions using a multidimensional Kolmogorov–Smirnov test. Monthly Notices of the Royal Astronomical Society , 451(3):2610–2624, 06 2015

    2015

  18. [26]

    Hoeffding

    W. Hoeffding. A Class of Statistics with Asymptotically Normal Distribution. The Annals of Mathematical Statistics , 19(3):293–325, 1948

    1948

  19. [27]

    B. Hoff. The Brownian Frame Process as a Rough Path . PhD thesis, University of Oxford, 2005

    2005

  20. [28]

    Horvath, M

    B. Horvath, M. Lemercier, C. Liu, T. Lyons, and C. Salvi. Optimal Stopping via Distribution Regression: a Higher Rank Signature Approach. 2023

    2023

  21. [29]

    Z. Issa, B. Horvath, M. Lemercier, and C. Salvi. Non-adversarial training of Neural SDEs with signature kernel scores. In Advances in Neural Information Processing Systems , volume 36, pages 11102–11126. Curran Associates, Inc., 2023

    2023

  22. [30]

    N. L. Johnson, S. Kotz, and N. Balakrishnan. Continuous Univariate Distributions , volume 1. John Wiley and Sons, 2nd edition, 1994

    1994

  23. [31]

    F. J. Kiraly and H. Oberhauser. Kernels for Sequentially Ordered Data. Journal of Machine Learning Research, 20(31):1–45, 2019

    2019

  24. [32]

    J. M. K¨ ubler, W. Jitkrittum, B. Sch¨ olkopf, and K. Muandet. A Witness Two-Sample Test. In Proceedings of The 25th International Conference on Artificial Intelligence and Statistics, volume 151 of Proceedings of Machine Learning Research (PMLR) , pages 1403–1419. PMLR, 28–30...

    2022

  25. [33]

    Lee and H

    D. Lee and H. Oberhauser. The Signature Kernel. 2023. 43

    2023

  26. [34]

    Lemercier, C

    M. Lemercier, C. Salvi, T. Damoulas, E. Bonilla, and T. Lyons. Distribution Regression for Sequential Data. In Proceedings of the 24th International Conference on Artificial Intelligence and Statistics , volume 130 of Proceedings of Machine Learning Research (PMLR), pages 3754...

    2021

  27. [35]

    Lyons and H

    T. Lyons and H. Ni. Expected signature of Brownian motion up to the first exit time from a domain. The Annals of probability , 43(5):2729–2762, 2015

    2015

  28. [36]

    Muandet, K

    K. Muandet, K. Fukumizu, B. Sriperumbudur, and B. Sch¨ olkopf.Kernel Mean Embed- ding of Distributions: A Review and Beyond . 2017

    2017

  29. [37]

    H. Ni, L. Szpruch, M. Sabate-Vidales, B. Xiao, M. Wiese, and S. Liao. Sig-wasserstein gans for time series generation. In Proceedings of the Second ACM International Con- ference on AI in Finance , ICAIF ’21, New York, NY, USA, 2022. Association for Com- puting Machinery

    2022

  30. [38]

    D. Pati. U-statistics

  31. [39]

    Rahimi and B

    A. Rahimi and B. Recht. Random Features for Large-Scale Kernel Machines. In Ad- vances in Neural Information Processing Systems , volume 20. Curran Associates, Inc., 2007

    2007

  32. [40]

    Salvi, T

    C. Salvi, T. Cass, J. Foster, T. Lyons, and W. Yang. The Signature Kernel is the Solution of a Goursat PDE. SIAM Journal on Mathematics of Data Science , 3(3):873–899, 2021

    2021

  33. [41]

    Salvi, M

    C. Salvi, M. Lemercier, C. Liu, B. Horvath, T. Damoulas, and T. Lyons. Higher Or- der Kernel Mean Embeddings to Capture Filtrations of Stochastic Processes. In 35th Conference on Neural Information Processing Systems , volume 34 of NeurIPS 2021 , 2021

    2021

  34. [42]

    Schrab, I

    A. Schrab, I. Kim, M. Albert, B. Laurent, B. Guedj, and A. Gretton. MMD Aggregated Two-Sample Test. Journal of machine learning research , 24(194):1–81, 2023

    2023

  35. [43]

    J. Shao. U- and v-statistics, 2018. 44

    2018

  36. [44]

    Simon-Gabriel and B

    C.-J. Simon-Gabriel and B. Sch¨ olkopf. Kernel Distribution Embeddings: Universal Ker- nels, Characteristic Kernels and Kernel Metrics on Distributions. Journal of Machine Learning Research, 19(44):1–29, 2018

    2018

  37. [45]

    D. J. Sutherland, H. Tung, H. Strathmann, S. De, A. Ramdas, A. J. Smola, and A. Gret- ton. Generative Models and Model Criticism via Optimized Maximum Mean Discrep- ancy. In 5th International Conference on Learning Representation , ICLR 2017, 2017

    2017

  38. [46]

    Terry and A

    L. Terry and A. D. McLeod. Signature Methods in Machine Learning. 2024

    2024

  39. [47]

    Rough paths, signatures and the modelling of functions on streams

    Terry Lyons. Rough paths, signatures and the modelling of functions on streams. In Proceedings of the International Congress of Mathematicians , 2014

    2014

  40. [48]

    T´ oth, H

    C. T´ oth, H. Oberhauser, and Z. Szab´ o. Random Fourier Signature Features. 2023

    2023

  41. [49]

    G. E. Uhlenbeck and L. S. Ornstein. On the Theory of Brownian Motion. Physics Review, 36(5):823–841, 1930

    1930

  42. [50]

    Zhang, X

    P. Zhang, X. Chen, L. Zhao, W. Xiong, T. Qin, and T.-Y. Liu. Distributional Rein- forcement Learning for Multi-Dimensional Reward Functions. In Advances in Neural Information Processing Systems, volume 34, pages 1519–1529. Curran Associates, Inc., 2021. 45

    2021

  43. [2022]

    King’s College London

Pith tools

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