Pith. sign in

REVIEW 3 major objections 6 minor 58 references

LSTM and Transformer networks recover fractional Hawkes parameters from inter-arrival sequences far more accurately than maximum likelihood, without ever evaluating the likelihood.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

Transformer and LSTM models recover FHP parameters (μ,γ,α,β) from inter-arrival sequences with lower MSE than classical MLE on synthetic data and pass distributional checks on AAPL and 911 data.

T0 review reviewed 2026-07-14 challenge →

load-bearing objection Usable first neural estimator for fractional Hawkes parameters; the headline MLE gap is almost certainly inflated by a broken classical baseline, but the synthetic maps and real-data predictive checks still look real. the 3 major comments →

arxiv 2607.11177 v1 pith:DHRXEKZM submitted 2026-07-13 cs.LG stat.ML

NeuroMem-FHP: A Likelihood-Free Deep Learning Framework for Parameter Estimation of Fractional Hawkes Process

Neha Gupta , Aditya Maheshwari This is my paper
classification cs.LG stat.ML MSC 60G5560G2262M0968T07
keywords fractional Hawkes processparameter estimationTransformerLSTMlikelihood-free inferencelong-range dependenceMittag-Leffler kernelpoint processes
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

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 introduces NeuroMem-FHP, a supervised deep-learning framework that maps fixed-length windows of inter-arrival times drawn from a fractional Hawkes process directly onto the four parameters that govern baseline rate, decay, branching, and long-memory strength. Classical maximum-likelihood estimation of the same Mittag-Leffler kernel is numerically costly and unstable; the networks avoid that cost entirely. On large synthetic test sets the Transformer reaches an MSE of 0.16 while stabilized MLE remains above 2.8; the same trained models, applied zero-shot to AAPL trade timestamps and Montgomery County 911 calls, generate simulated sequences whose empirical distribution, heavy tails and autocorrelation structure match the real data. A reader who works with high-frequency event streams therefore obtains a practical, likelihood-free route to long-memory point-process modeling.

Core claim

When trained on synthetic trajectories of the fractional Hawkes process, both an LSTM and a Transformer learn a direct map from observed inter-arrival windows to the constrained parameter vector (μ, γ, α, β) whose mean-squared error is an order of magnitude lower than that of numerically stabilized maximum-likelihood estimation; the Transformer-estimated parameters further reproduce the ECDF, survival function and ACF of two real high-frequency event series.

What carries the argument

NeuroMem-FHP: a supervised neural regression that ingests windowed inter-arrival sequences, encodes them with either gated recurrence or multi-head self-attention, and emits the four FHP parameters under softplus/sigmoid constraints, trained by MSE on thinning-simulated paths.

Load-bearing premise

The numerical comparison treats a carefully clipped and filtered maximum-likelihood optimizer as a fair classical baseline rather than an optimizer that systematically fails on the Mittag-Leffler likelihood.

What would settle it

Re-optimize the same synthetic 100-event windows with a multi-start or gradient-based MLE that converges on every replicate; if that MLE’s MSE then falls near the neural values of 0.16–0.18, the claimed superiority disappears.

Watch this falsifier. Get emailed when new claim-graph text bears on it.

If this is right

  • High-frequency event series with long-range dependence can be fitted by a single forward pass of a pre-trained Transformer instead of iterative likelihood maximization.
  • Once parameters are obtained, realistic synthetic catalogs that preserve empirical tails and autocorrelation become available for stress-testing or scenario generation.
  • Window length can be treated as a tunable hyper-parameter that trades estimation bias against variance of the recovered memory exponent β.
  • The same architecture supplies a template for other non-Markovian point processes whose likelihoods are intractable.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The dramatic MLE gap may largely reflect numerical fragility of the Mittag-Leffler likelihood rather than pure representational superiority, so a hybrid that uses the network only as a warm start could close the remaining error.
  • Zero-shot transfer succeeds only when real data lie inside the synthetic parameter box and share the same log-transform and truncation preprocessing.
  • Extending the output head and simulator to multivariate or marked fractional Hawkes processes would immediately yield a general likelihood-free engine for multi-type event streams.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 6 minor

Summary. The paper proposes NeuroMem-FHP, a likelihood-free supervised framework that maps fixed-length sequences of inter-arrival times from a fractional Hawkes process (Mittag-Leffler kernel) to the four parameters (μ, γ, α, β) using an LSTM and a Transformer. Synthetic paths are generated by thinning over broad uniform parameter ranges; networks are trained by MSE against ground-truth parameters with positivity/interval constraints on the outputs. On held-out synthetic data the authors report substantially lower MSE than a numerically stabilized MLE baseline (Transformer 0.1634 / LSTM 0.1752 vs MLE 2.8032 in the abstract and §4.1; Table 1 gives related figures). Ablations examine batch size, epochs, hidden size and sample size. On AAPL NBBO and Montgomery County 911 inter-arrival windows the trained Transformer produces parameter estimates from which simulated sequences approximately match empirical ECDF, survival tails and ACF; a window-size sensitivity study is also reported.

Significance. If the synthetic superiority claim is robust, the work supplies a practical, likelihood-free estimator for a non-Markovian self-exciting process whose Mittag-Leffler likelihood is known to be numerically delicate, and it is among the first deep-learning treatments of fractional Hawkes parameter recovery. The real-data predictive checks (ECDF/survival/ACF match under the maintained FHP model) and the window-size analysis are useful applied contributions for high-frequency event modeling. The contribution is primarily empirical and engineering-oriented rather than theoretical; its lasting value hinges on whether the MLE baseline is a competent classical estimator and on clearer reporting of absolute recovery quality (R², per-parameter errors).

major comments (3)
  1. The headline claim that both neural models “significantly outperform classical MLE” (§4.1, abstract, Figure 3, Table 1) rests on a single numerically stabilized MLE implementation whose reported MSE of ~2.8–2.9 is implausibly large for a four-dimensional parameter vector whose components mostly lie in (0,2]. The paper retains only convergent runs after clipping and log stabilization, but supplies no multi-start protocol, no Mittag-Leffler evaluation accuracy, no fraction of discarded trajectories, and no recovery diagnostics on the same synthetic draws used for the nets. An MSE of that magnitude is the signature of optimizer collapse on the non-Markovian likelihood (Eq. 3) rather than of an informative baseline. Without a competent classical comparator (or an explicit diagnosis of MLE failure rates), the claimed gap cannot be interpreted as intrinsic superiority of the neural map.
  2. Ablation metrics in §4.2 / Figure 4 show R² peaking near ~0.28 (and remaining in the 0.22–0.28 range across batch size, epochs, hidden size and sample size). That implies most of the parameter variance remains unexplained even while MSE is framed as low. The abstract and §4.1 lead with MSE alone and do not reconcile the low R² with the claim of accurate recovery of (μ, γ, α, β). Per-parameter bias/variance or R² (especially for the fractional memory parameter β) should be reported so that absolute identification quality can be assessed.
  3. Internal inconsistencies in experimental protocol undermine reproducibility of the synthetic results. Sequence length is stated as T=50 in the Transformer architecture description (§3.2) but T=100 in data generation and results (§3.3, §4.1); γ ranges differ between §3.3 (U(0.1,2.0)) and §4.1 ([0.10,5.0]); Table 1 MSE numbers (0.117 / 0.140 / 2.901) do not match the abstract/§4.1 figures (0.1634 / 0.1752 / 2.8032); and the LSTM description oscillates between a single FC layer and an intermediate ReLU FC layer. These discrepancies must be resolved and a single, fully specified protocol fixed before the quantitative claims can be trusted.
minor comments (6)
  1. Abstract and opening sentence contain grammatical errors (“Experiments on synthetic data that both neural models…”; “The proposed framework is also on two real-world…”).
  2. §3.1 contains duplicated and contradictory statements of the output activations (Softplus/Sigmoid ranges and the 0.9 scaling appear twice with slightly different wording).
  3. Figure 3 caption and surrounding text should state whether MSE is averaged over the four parameters, over samples, or both, and whether it is computed on the constrained or unconstrained outputs.
  4. Real-data preprocessing (log10 transform, exclusion of gaps >100 s) is reasonable but should be justified against the synthetic training distribution so that zero-shot transfer assumptions are explicit.
  5. Table 1 lists R² in the prose description but the table columns do not include an R² column; align text and table.
  6. Several references and arXiv preprints are recent and relevant; a short comparison to other likelihood-free or neural Hawkes estimators would help position the contribution.

Circularity Check

0 steps flagged

No circularity: synthetic MSE evaluates against independently known ground-truth parameters; real-data checks are ordinary simulate-from-fit validation under the maintained FHP model.

full rationale

The paper's central claims rest on two non-circular evaluations. On synthetic data (Section 4.1), FHP trajectories are generated by the thinning algorithm with parameters drawn from known uniform ranges; the LSTM/Transformer are trained by supervised MSE regression to recover those parameters from inter-arrival windows, and test MSE (0.1634/0.1752 vs MLE 2.8032) is computed against the same known ground truth. Nothing is fitted then re-predicted. On real AAPL NBBO and MontcoAlert 911 data (Section 5), the already-trained network produces parameter estimates that are used only to resimulate FHP paths; ECDF, survival, and ACF agreement is a standard predictive check under the model assumption, not a definitional identity or a fitted quantity renamed as a prediction. No equation reduces a reported result to its own inputs by construction. The single self-citation (arXiv:2512.05893 on a related fractional Poisson estimator) appears only as related work and is not invoked as a uniqueness theorem or load-bearing premise. The weak MLE baseline is a correctness concern, not circularity. The derivation chain is therefore self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

5 free parameters · 4 axioms · 1 invented entities

The central claim rests on the standard FHP intensity with Mittag-Leffler kernel, correctness of Ogata thinning for data generation, and the modeling choice that fixed-length inter-arrival windows plus MSE regression suffice to identify four constrained parameters. Network and simulation hyperparameters are free design choices; NeuroMem-FHP is branding rather than a new physical entity. Real-data conclusions further assume the empirical series are well described by univariate FHP after ad hoc cleaning (drop zeros/negatives, drop gaps >100s, log transforms).

free parameters (5)
  • Training parameter ranges (μ,γ,α,β uniforms) = μ~U(0.1,2); γ~U(0.1,2) or U(0.1,5); α,β~U(0.01,0.9)
    Synthetic labels are drawn from hand-chosen boxes (e.g. μ∈[0.1,2], γ∈[0.1,2] or [0.1,5], α,β∈[0.01,0.9]); these bounds define the supervised task and transfer regime.
  • Input window length T = 50 or 100 (inconsistent)
    Fixed sequence length used as network input; paper states both T=50 and T=100 in different sections; real-data sensitivity uses W∈{20..200}.
  • Network capacity and training hyperparameters = hidden/d_model=32; lr=1e-3; ~100 epochs
    d_model/hidden=32, L=1 encoder, H=2 heads, Adam lr=1e-3, epochs≈100, batch sizes in ablation; chosen to fit the regression task.
  • Output constraint scalings (softplus offset, 0.9·sigmoid) = softplus+0.01; 0.9*sigmoid
    Hard-coded maps force μ,γ>0 and α,β∈(0,0.9) rather than full theoretical (0,1); affects admissible estimates.
  • Real-data cleaning thresholds = exclude Δt>100s; log(1+Δt) or log10
    Drop inter-arrivals >100s and apply log/log10 transforms before inference; changes empirical distribution fed to the estimator.
axioms (4)
  • domain assumption FHP conditional intensity uses Mittag-Leffler kernel f_{β,γ}(t)=γ t^{β-1} E_{β,β}(-γ t^β) with parameters (μ,γ,α,β) as in §2.2–2.3.
    All simulation labels and the estimation target are defined by this model class.
  • standard math Ogata thinning with intensity upper bound M correctly samples the FHP used for supervised training (§3.3).
    Training data quality depends on correct simulation of the target process.
  • ad hoc to paper A fixed-length inter-arrival window contains enough information for a neural map to identify all four parameters under MSE loss.
    Identifiability is assumed via architecture choice; low ablation R² suggests partial failure of this premise.
  • domain assumption After preprocessing, AAPL NBBO and MontcoAlert 911 series are adequately described as univariate FHP for predictive validation (§5).
    Real-data conclusions require the model class to be appropriate, which is not independently tested against alternatives.
invented entities (1)
  • NeuroMem-FHP framework no independent evidence
    purpose: Brand name for applying standard LSTM/Transformer regressors to FHP inter-arrival sequences.
    No new mathematical object beyond existing FHP plus off-the-shelf sequence models; independent_evidence false as a distinct entity.

reviewed 2026-07-14 · how reviews work

0 comments
Cite this review

Pith. "Pith review of NeuroMem-FHP: A Likelihood-Free Deep Learning Framework for Parameter Estimation of Fractional Hawkes Process." pith.science (2026). https://pith.science/paper/DHRXEKZM

@misc{pith2026260711177,
  author       = {Pith},
  title        = {Pith review of: NeuroMem-FHP: A Likelihood-Free Deep Learning Framework for Parameter Estimation of Fractional Hawkes Process},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/DHRXEKZM}},
  note         = {Machine review of arXiv:2607.11177}
}
Share X Bluesky LinkedIn Reddit HN
abstract

In this paper, we propose deep learning based NeuroMem-FHP framework for estimating the parameters of the fractional Hawkes process (FHP), a self-exciting point process that captures long-range dependence through a fractional Mittag-Leffler excitation kernel. Two neural architectures, namely a Long Short-Term Memory (LSTM) network and a Transformer, are developed to estimate the model parameters $(\mu,\gamma,\alpha,\beta)$ directly from sequences of inter-arrival times without requiring computationally intensive likelihood optimization. Experiments on synthetic data that both neural models significantly outperform the classical Maximum Likelihood Estimation (MLE) method, with the Transformer achieving the highest estimation accuracy (MSE = $0.1634$), followed by the LSTM (MSE = $0.1752$), compared to MLE (MSE = $2.8032$). An ablation study further examines the effects of key hyperparameters on model performance. The proposed framework is also on two real-world high-frequency datasets, namely AAPL NBBO transaction data and Montgomery County 911 emergency call records. Using a predictive validation approach, event sequences simulated from the estimated parameters closely reproduce the empirical distribution, tail behavior, and temporal dependence structure of the observed data. These results demonstrate that Transformer-based parameter estimation provides an accurate and efficient alternative to conventional estimation techniques for FHP and offers a promising framework for modeling event-driven systems with long-memory dynamics.

Figures

Figures reproduced from arXiv: 2607.11177 by Aditya Maheshwari, Neha Gupta.

Figure 1
Figure 1. Figure 1: LSTM-based architectures for parameter estimation of the FHP: µ > 0, γ > 0, α ∈ (0, 1) and β ∈ (0, 1). 5 [PITH_FULL_IMAGE:figures/full_fig_p005_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Transformer-based architectures for parameter estimation of the FHP: µ > 0, γ > 0, α ∈ (0, 1) and β ∈ (0, 1). 7 [PITH_FULL_IMAGE:figures/full_fig_p007_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Comparison of parameter estimation accuracy for the FHP using RNN￾LSTM, Transformer, and MLE. The deep learning models significantly outper￾form the classical MLE approach, with the Transformer achieving the lowest Mean Squared Error. 4.2. Ablation Studies. Here, we analyzed several results to demonstrate the performance of the Transformer model for parameter estimation of the FHP. These results include th… view at source ↗
Figure 4
Figure 4. Figure 4: Ablation study of the Transformer-based FHP model. The plots illus￾trate the impact of batch size, number of epochs, hidden layer size, and training sample size on score metrics. The top-right plot [PITH_FULL_IMAGE:figures/full_fig_p011_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Performance evaluation of the Transformer-based FHP framework: (a) ECDF comparison between real (AAPL) and simulated inter-arrival times. (b) Survival probability comparison demonstrating the ability of the model to reproduce the tail behavior of the empirical distribution. (c) ACF comparison showing that the simulated sequence preserves the dependence structure observed in the real data (AAPL). (d) Traini… view at source ↗
Figure 6
Figure 6. Figure 6: Performance evaluation of the proposed Transformer-based FHP on the 911 dataset. (a) ECDF comparison between real and simulated inter-arrival times. (b) Survival probability comparison demonstrating the ability of the model to repro￾duce the tail behavior of the empirical distribution. (c) ACF comparison showing that the simulated sequence preserves the dependence structure observed in the real data. (d) T… view at source ↗
Figure 7
Figure 7. Figure 7: Statistical analysis of the AAPL event sequence: (a) autocorrelation function (ACF), (b) bias as a function of window size, and (c) variance as a function of window size [PITH_FULL_IMAGE:figures/full_fig_p016_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Statistical analysis of the 911 event sequence: (a) autocorrelation func￾tion (ACF), (b) bias as a function of window size, and (c) variance as a function of window size. whether the temporal dependence structure is preserved, we compare the empirical ACF of the observed event sequences with the ACF obtained from Transformer-generated simulations across different window sizes. The simulated ACFs closely fo… view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

58 extracted references · 4 linked inside Pith

  1. [1]

    J. A. Achcar, E. R. Rodrigues, C. D. Paulino, and P. Soares. Non-homogeneous poisson models with a change- point: an application to ozone peaks in mexico city.Environ. Ecol. Stat., 17(4):521–541, Dec. 2010

    2010

  2. [2]

    J. L. Ba, J. R. Kiros, and G. E. Hinton. Layer normalization.arXiv preprint arXiv:1607.06450, 2016

    Pith/arXiv arXiv 2016

  3. [3]

    Bacry, I

    E. Bacry, I. Mastromatteo, and J.-F. Muzy. Hawkes processes in finance.Market Microstructure and Liquidity, 01(01):1550005, 2015

    2015

  4. [4]

    Bauwens and N

    L. Bauwens and N. Hautsch.Modelling Financial High Frequency Data Using Point Processes, pages 953–979. Springer Berlin Heidelberg, Berlin, Heidelberg, 2009

    2009

  5. [5]

    Boros, B

    D. Boros, B. Csan´ ady, I. Ivkovic, L. Nagy, A. Luk´ acs, and L. M´ arkus. Deep learning the hurst parameter of linear fractional processes and assessing its reliability.Quality and Reliability Engineering International, 40(8):4228– 4246, 2024

    2024

  6. [6]

    C. G. Bowsher. Modelling security market events in continuous time: Intensity based, multivariate point process models.Journal of Econometrics, 141(2):876–912, 2007

    2007

  7. [7]

    Cartea, S

    ´A. Cartea, S. Jaimungal, and J. Ricci. Buy low, sell high: A high frequency trading perspective.SIAM Journal on Financial Mathematics, 5(1):415–444, 2014

    2014

  8. [8]

    Chavez-Demoulin, A

    V. Chavez-Demoulin, A. C. Davison, and A. J. McNeil. Estimating value-at-risk: a point process approach. Quantitative Finance, 5(2):227–234, 2005

    2005

  9. [9]

    J. Chen, A. G. Hawkes, and E. Scalas. A fractional hawkes process. InNonlocal and fractional operators, pages 121–131. Springer, 2021

    2021

  10. [10]

    Cranmer, A

    M. Cranmer, A. Sanchez Gonzalez, P. Battaglia, R. Xu, K. Cranmer, D. Spergel, and S. Ho. Discovering symbolic models from deep learning with inductive biases.Advances in neural information processing systems, 33:17429– 17442, 2020

    2020

  11. [11]

    Csan´ ady, L

    B. Csan´ ady, L. Nagy, D. Boros, I. Ivkovic, D. Kov´ acs, D. T´ oth-Lakits, L. M´ arkus, and A. Luk´ acs. Parameter estimation of long memory stochastic processes with deep neural networks.arXiv preprint arXiv:2410.03776, 2024

    Pith/arXiv arXiv 2024

  12. [12]

    Davis, B

    L. Davis, B. Baeumer, and T. Wang. A fractional hawkes process model for earthquake aftershock sequences. Journal of the Royal Statistical Society Series C: Applied Statistics, 73(5):1185–1202, 2024

    2024

  13. [13]

    Z. Ding, C. W. Granger, and R. F. Engle. A long memory property of stock market returns and a new model. Journal of Empirical Finance, 1(1):83 – 106, 1993

    1993

  14. [14]

    Doukhan, G

    P. Doukhan, G. Oppenheim, and M. S. Taqqu, editors.Theory and applications of long-range dependence. Birkh¨ auser Boston, Inc., Boston, MA, 2003

    2003

  15. [15]

    Dupret and D

    J.-L. Dupret and D. Hainaut. A fractional hawkes process for illiquidity modeling.Mathematics and Financial Economics, 19(1):143–181, 2025

    2025

  16. [16]

    Fein-Ashley

    J. Fein-Ashley. A comparison of traditional and deep learning methods for parameter estimation of the ornstein- uhlenbeck process.arXiv preprint arXiv:2404.11526, 2024

    Pith/arXiv arXiv 2024

  17. [17]

    J. Feng, X. Wang, Q. Liu, Y. Li, and Y. Xu. Deep learning-based parameter estimation of stochastic differential equations driven by fractional brownian motions with measurement noise.Communications in Nonlinear Science and Numerical Simulation, 127:107589, 2023

    2023

  18. [18]

    Goodfellow, Y

    I. Goodfellow, Y. Bengio, and A. Courville.Deep Learning. MIT Press, 2016

    2016

  19. [19]

    Greff, R

    K. Greff, R. K. Srivastava, J. Koutn´ ık, B. R. Steunebrink, and J. Schmidhuber. Lstm: A search space odyssey. IEEE transactions on neural networks and learning systems, 28(10):2222–2232, 2016

    2016

  20. [20]

    Gupta and A

    N. Gupta and A. Maheshwari. Neuromemfpp: A recurrent neural approach for memory-aware parameter esti- mation in fractional poisson process.arXiv preprint arXiv:2512.05893, 2025

    arXiv 2025

  21. [21]

    Habyarimana, J

    C. Habyarimana, J. A. Aduda, and E. Scalas. Parameter estimation for the fractional hawkes process.Journal of Agricultural, Biological and Environmental Statistics, pages 1–26, 2024

    2024

  22. [22]

    Habyarimana, J

    C. Habyarimana, J. A. Aduda, E. Scalas, J. Chen, A. G. Hawkes, and F. Polito. A fractional hawkes process ii: Further characterization of the process.Physica A: Statistical Mechanics and its Applications, 615:128596, 2023. 17

    2023

  23. [23]

    D. Hainaut. Clustered L´ evy processes and their financial applications.Journal of computational and applied mathematics, 319:117–140, 2017

    2017

  24. [24]

    D. Hainaut. Fractional hawkes processes.Physica A: Statistical Mechanics and its Applications, 549:124330, 2020

    2020

  25. [25]

    Hainaut and F

    D. Hainaut and F. Moraux. A switching self-exciting jump diffusion process for stock prices.Ann. Fin., 15(2):267– 306, June 2019

    2019

  26. [26]

    A. G. Hawkes. Point spectra of some mutually exciting point processes.Journal of the Royal Statistical Society Series B: Statistical Methodology, 33(3):438–443, 1971

    1971

  27. [27]

    A. G. Hawkes. Spectra of some self-exciting and mutually exciting point processes.Biometrika, 58(1):83–90, 1971

    1971

  28. [28]

    A. G. Hawkes. Cluster models for earthquakes-regional comparisons.Bull. Int. Stat. Inst., 45(3):454–461, 1973

    1973

  29. [29]

    K. He, X. Zhang, S. Ren, and J. Sun. Deep residual learning for image recognition. InProceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pages 770–778, 2016

    2016

  30. [30]

    Hochreiter and J

    S. Hochreiter and J. Schmidhuber. Long short-term memory.Neural computation, 9(8):1735–1780, 1997

    1997

  31. [31]

    Karagiannis, M

    T. Karagiannis, M. Molle, and M. Faloutsos. Long-range dependence ten years of internet traffic modeling. Internet Computing, IEEE, 8(5):57–64, 2004

    2004

  32. [32]

    Kerss, N

    A. Kerss, N. N. Leonenko, and A. Sikorskii. Fractional Poisson fields and martingales.Journal of Applied Prob- ability, 56(3):868–887, 2019

    2019

  33. [33]

    D. P. Kingma and J. Ba. Adam: A method for stochastic optimization.arXiv preprint arXiv:1412.6980, 2014

    Pith/arXiv arXiv 2014

  34. [34]

    Lawrence, S

    E. Lawrence, S. V. Wiel, C. Law, S. B. Spolaor, and G. C. Bower. The nonhomogeneous poisson process for fast radio burst rates.The Astronomical Journal, 154(3):117, aug 2017

    2017

  35. [35]

    K. Lee. Recurrent neural network based parameter estimation of hawkes model on high-frequency financial data. Finance Research Letters, 55:103922, 2023

    2023

  36. [36]

    S. Li, Y. Ruan, C. Long, and Y. Cheng. Efficient cnn-lstm based parameter estimation of l´ evy driven stochastic differential equations. In2023 International Conference on Machine Learning and Applications (ICMLA), pages 316–323. IEEE, 2023

    2023

  37. [37]

    S. Li, Y. Wu, and Y. Cheng. Parameter estimation and random number generation for student l´ evy processes. Computational Statistics & Data Analysis, 194:107933, 2024

    2024

  38. [38]

    Lim and S

    B. Lim and S. Zohren. Time-series forecasting with deep learning: a survey.Philosophical transactions of the royal society a: mathematical, physical and engineering sciences, 379(2194), 2021

    2021

  39. [39]

    F. Liu, J. Li, and L. Wang. Pi-lstm: Physics-informed long short-term memory network for structural response modeling.Engineering Structures, 292:116500, 2023

    2023

  40. [40]

    P. Lu, L. Qiu, W. Yu, S. Welleck, and K.-W. Chang. A survey of deep learning for mathematical reasoning. In Proceedings of the 61st annual meeting of the association for computational linguistics (volume 1: long papers), pages 14605–14631, 2023

    2023

  41. [41]

    Mahar, S

    N. Mahar, S. Sen, and L. Mevel. An attention-augmented long short-term memory network for pinn-based structural health monitoring.Engineering Structures, 340:120673, 2025

    2025

  42. [42]

    K. D. Majeske. A non-homogeneous poisson process predictive model for automobile warranty claims.Reliability Engineering & System Safety, 92(2):243–251, 2007

    2007

  43. [43]

    Nair and G

    V. Nair and G. E. Hinton. Rectified linear units improve restricted Boltzmann machines. InProceedings of the 27th International Conference on Machine Learning, pages 807–814, 2010

    2010

  44. [44]

    Y. Ogata. On Lewis’ simulation method for point processes.IEEE Transactions on Information Theory, 27(1):23– 31, 1981

    1981

  45. [45]

    Y. Ogata. Statistical models for earthquake occurrences and residual analysis for point processes.Journal of the American Statistical association, 83(401):9–27, 1988

    1988

  46. [46]

    Y. Ogata. Space-time point-process models for earthquake occurrences.Annals of the Institute of Statistical Mathematics, 50:379–402, 1998

    1998

  47. [47]

    ¨Ozalp, G

    E. ¨Ozalp, G. Margazoglou, and L. Magri. Physics-informed long short-term memory for forecasting and recon- struction of chaos. InInternational Conference on Computational Science, pages 382–389. Springer, 2023

    2023

  48. [48]

    A. Pagan. The econometrics of financial markets.Journal of Empirical Finance, 3(1):15 – 102, 1996

    1996

  49. [49]

    Pawar, R

    K. Pawar, R. S. Jalem, and V. Tiwari. Stock market price prediction using lstm rnn. InEmerging Trends in Expert Applications and Security: Proceedings of ICETEAS 2018, pages 493–503. Springer, 2018

    2018

  50. [50]

    M. D. Porter and G. White. Self-exciting hurdle models for terrorist activity.The Annals of Applied Statistics, 6(1):106 – 124, 2012

    2012

  51. [51]

    Sagheer and M

    A. Sagheer and M. Kotb. Time series forecasting of petroleum production using deep lstm recurrent networks. Neurocomputing, 323:203–213, 2019

    2019

  52. [52]

    D. Szarek. Neural network-based anomalous diffusion parameter estimation approaches for gaussian processes. International Journal of Advances in Engineering Sciences and Applied Mathematics, 13(2):257–269, 2021. 18

    2021

  53. [53]

    Temirkhanov, K

    A. Temirkhanov, K. Kuvshinova, and D. Simakov. An in-context approach to estimate the parameters of a stochastic process. InAI4X 2025 International Conference

    2025

  54. [54]

    Varotsos and D

    C. Varotsos and D. Kirk-Davidoff. Long-memory processes in ozone and temperature variations at the region 60 ◦s-60 ◦n.Atmospheric Chemistry and Physics, 6(12):4093–4100, 2006

    2006

  55. [55]

    Vaswani, N

    A. Vaswani, N. Shazeer, N. Parmar, J. Uszkoreit, L. Jones, A. N. Gomez, L. Kaiser, and I. Polosukhin. Attention is all you need.Advances in Neural Information Processing Systems, 30, 2017

    2017

  56. [56]

    X. Wang, J. Feng, Q. Liu, Y. Li, and Y. Xu. Neural network-based parameter estimation of stochastic differential equations driven by l´ evy noise.Physica A: Statistical Mechanics and its Applications, 606:128146, 2022

    2022

  57. [57]

    Wang and Z.-X

    X.-T. Wang and Z.-X. Wen. Poisson fractional processes.Chaos Solitons Fractals, 18(1):169–177, 2003

    2003

  58. [58]

    Yin and D

    X. Yin and D. S. Yin. Transformer-based parameter estimation in statistics.Mathematics, 12(7):1040, 2024. 19

    2024

This paper was first reviewed by grok-4.5 on July 14, 2026.