Pith. sign in

REVIEW 2 major objections 3 minor 48 references

Nonparametric Filtering, Estimation and Classification using Neural Jump ODEs

T0 review · 2 major / 3 minor · reviewed 2026-08-11 · deepseek-v4-flash

Pith's one-line read Neural jump ODEs provably learn the optimal online filter

desk verdict The IO NJODE extension is a genuine, mostly careful contribution, but Theorem 4.4 as proved does not cover the paper's own threshold-classification example, so the convergence claims need reworking before I'd rely on them. read the letter →

arxiv 2412.03271 v2 pith:24SMSRWH submitted 2024-12-04 stat.ML cs.LGcs.NAmath.NAmath.OCmath.PR

classification stat.MLcs.LGcs.NAmath.NAmath.OCmath.PR MSC 62M4560G3562G0568T07
keywords neuraljumpODEonlinefilteringinput-outputsystemsconditionalexpectationirregularobservationssignaturetransformL2-optimalclassification
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 extends the Neural Jump ODE framework to input-output systems, where the observed input process and the target output process are distinct. It introduces the IO NJODE model, which evolves a hidden state continuously between observations and jumps at observation times, together with a new objective that squares the pre-jump and post-jump errors separately. The main result is that minimizers of this objective converge to the true conditional expectation as the network size and signature truncation grow, making the model a provably L2-optimal online filter. The practical payoff is that filtering, parameter estimation, and classification become fully data-driven: only sample paths with irregular and partially masked observations are needed, not the underlying distributions.

What carries the argument

The load-bearing object is the truncated signature of a piecewise-linear interpolation of the observed input path, fed through bounded-output feedforward neural networks inside a jump ODE: the hidden state follows an ODE between observation times and is reset at observation times. The new loss (4), which squares the error just before an observation and the error at the observation separately, is what makes the argument work. Lemma 4.2 decomposes this loss into the intrinsic filter error plus the squared distance from the model to $\hat V$, proving uniqueness of the conditional expectation as the minimizer; Lemma 4.3 converts sums over random observation times into expectations over a uniform random variable, allowing the proof to treat observation times as an independent source of randomness; and the signature approximation result lets continuous functions of bounded-variation paths be approximated uniformly by functions of the truncated signature.

What would settle it

Take an input-output system where the optimal filter is discontinuous in time, for example an output that jumps at a random time observed with positive probability, and train a large IO NJODE; if the model's loss stays strictly above the conditional expectation's loss as network size grows, the convergence claim fails because Assumptions 2 and 3 are violated.

Watch

Extended reading notes

Core claim

The central claim is that, for an observed input process $U$ and a target output process $V$, the IO NJODE trained with the objective function (4) converges to the conditional expectation $\hat V_t = \mathbb{E}[V_t\mid A_t]$ in the pseudo-metric $d_k$ as the neural network size and signature truncation level tend to infinity. Theorem 4.4 establishes this for the theoretical loss, and Theorem 4.5 shows that the same conclusion holds when the loss is replaced by its Monte Carlo approximation on finitely many training paths, with an almost-sure subsequence of trained models converging to the optimal filter. The proof builds on the fact that the new loss has a Pythagorean decomposition around the conditional expectation, so $\hat V$ is its unique minimizer among adapted processes, and on universal approximation results for truncated signatures and neural networks.

Load-bearing premise

The optimal filter's conditional-expectation functions must be continuously differentiable in time, so the proof gives no guarantee for filters that jump when information arrives.

Editorial extensions

If this is right

  • For any input-output system satisfying the assumptions, the trained IO NJODE is an L2-optimal online filter: in the limit its output is indistinguishable from the conditional expectation at observation times.
  • Because training needs only sample paths rather than the underlying distributions, the method applies where classical particle filters are unusable or numerically unstable, including the CIR examples reported in the paper.
  • Choosing the output process to be indicator functions of class labels turns the same model into an online classifier of conditional class probabilities.
  • When all output coordinates also appear as input coordinates, the new loss reproduces the original NJODE setting, and the paper argues the earlier coupled loss is then preferable for learning jumps quickly.
  • With the Monte Carlo loss, convergence survives finite-sample training: as the sample size and then the model size grow, trained models converge almost surely to the optimal filter.

Reading between the lines

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

  • As an editorial extension, the same proof strategy should transfer to noisy observations and long-horizon predictions, since the paper states those cases only require replacing the objective and the metric.
  • A finite-sample trade-off follows: the squared-separately objective is the correct one for L2 optimality, but the old coupled objective learns jumps faster, so practitioners with limited data face a real design choice rather than a free lunch.
  • If the convergence is correct, online classifiers built by setting the output to class-indicator processes are L2-consistent estimators of class probabilities, a consequence the paper states but does not isolate as a separate theorem.
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 / 3 minor

Summary. The paper extends Neural Jump ODEs to input-output systems, where the input process U is observed (possibly irregularly and partially) and the output process V is to be estimated online. The main theoretical result, Theorem 4.4, states that under Assumptions 1–5 the minimizers of the new objective function (4) over growing network classes converge, in the pseudo-metrics d_k, to the true conditional expectation Vhat_t = E[V_t | A_t]; Theorem 4.5 extends this to the Monte Carlo objective. The authors introduce a squared per-term objective to make Vhat the unique minimizer in the input-output setting, prove uniqueness and approximation via truncated signatures and bounded-output neural networks, and provide examples in parameter filtering, stochastic filtering, and online classification. Experiments on synthetic data compare the method with particle filters and classical financial estimators, and Section 7 discusses why the old NJODE objective is inappropriate for input-output filtering.

Significance. If the convergence theorem is valid under the stated assumptions, the paper is a solid contribution: it gives a principled objective for online filtering in a nonparametric, data-driven setting, with a uniqueness result, a layered universal approximation argument, a Monte Carlo convergence theorem, and an open-source implementation with reproducible experiments. The explicit identification of the L2-optimal limit, Vhat, and the comparison with the old objective in Section 7 are valuable. However, the claimed scope currently exceeds what the proofs support: the online classification example in Section 5.3 does not satisfy Assumption 3 as used in the proof, and the uniqueness step in Theorem 4.4 uses an independence property of the observation masks that is not stated among the assumptions. Both issues are local and repairable, but they affect load-bearing parts of the central claim.

major comments (2)
  1. [Section 5.3 / Example 5.5 and Theorem 4.4, Step 2] The classification example does not satisfy Assumption 3, so the convergence theorem as stated does not cover it. For V_t = 1_{W_t >= alpha}, the natural version of the filter map is F(t,tau,w) = 1 - Phi((alpha - w)/sqrt(t - tau)) for t > tau. This function has no jointly continuous extension to the diagonal: along t downarrow tau with w = alpha it tends to 1 - Phi(0) = 1/2, whereas the diagonal value corresponding to the information at the observation time is 1_{w >= alpha}. The proof of Step 2 requires joint continuity on [0,T]^2 x P_epsilon to apply Proposition A.5. Moreover, the derivative f = partial_t F = phi((alpha - w)/sqrt(t - tau)) (alpha - w) / (2 (t - tau)^{3/2}) satisfies E|f(t)|^2 ~ c (t - tau)^{-3/2} when observation gaps can be arbitrarily small, so the integrability condition in Eq. (1) fails unless one imposes a positive minimum gap. The statement in Section 5.3 that Assumption 3 is easily verified for Example 5.5 is therefore not correct, and the claimed convergence for online classification is not proved as written.
  2. [Proof of Theorem 4.4, Step 1 (Eq. (9))] The proof factors E[1_{k<=n} M_{k,j} (Vhat_{t_k-} - eta_{t_k-})] as E[M_{k,j}] E[1_{k<=n}(Vhat_{t_k-} - eta_{t_k-})], which requires M_k to be independent of A_{t_k-}. Assumption 1 only says that M_k is independent of the observation times and of n; Assumption 6 says that the whole observation framework is independent of Z. Neither condition implies that M_k is independent of the earlier masks M_i (i<k), which are part of A_{t_k-}. If the mask process is serially dependent, the lower bound in Eq. (9) can fail, and the minimizer of the weighted objective Psi need not coincide with Vhat in the unweighted metric d_k. This is load-bearing for the uniqueness argument and for the final d_k-convergence step; please add an explicit independence or conditional independence condition on the masks, or modify the proof accordingly.
minor comments (3)
  1. [Theorem 4.5 and its proof] The statement and Step 3 use 'for every 1 <= k <= K', but K is never defined; the corresponding range in Definition 2.4 and Theorem 4.4 is 1 <= k <= bar n.
  2. [Eq. (12) and the surrounding text] The push-forward map mu_2 has coordinates (bar t, pi_m0, U_0, U*_t, n_t, delta_t), so its image lies in [0,T] x R^{d'} x R^{dU} x [0,1/epsilon]^2 x [epsilon,T]; the displayed set D_2 = [0,T]^2 x ... has one dimension too many and should be corrected.
  3. [Proof of Theorem 4.4, after Eq. (14)] The text refers to 'Assumption 2.1 on delta_min', but the relevant statement is Remark 2.1; also, the phrase '4th (in)equality' in Eq. (15) should be 'fourth equality'.

Circularity Check

0 steps flagged · score 2.0 of 10

No circular reduction found: V̂ is defined independently as E[V|A], the new loss (4) is proven (not assumed) to single it out, and the cited NJODE/signature results are prior published work used as real evidence.

full rationale

The central claim is not circular. The target V̂_t := E[V_t | A_t] is defined independently in Section 2.3 as a conditional expectation, and the new objective (4) is introduced as a squared loss whose minimizer is then proven, in Lemma 4.2 and in Step 1 of Theorem 4.4, to be exactly V̂. That proof uses the standard L2-projection property of conditional expectations together with Assumptions 1–5; it does not fit any parameter to V̂ and then rename the fit as a prediction. The signature-approximation step in Theorem 4.4 (Step 2) invokes Proposition A.5, which is cited to Krach et al. (2022, Prop. 3.8) and ultimately to classical signature results; those are prior published results with proofs, not assumptions containing the target conclusion, so citing them is self-citation but not circularity. Similarly, Lemmas 4.6 and 4.7 are prior lemmas from Krach et al. (2022), and the proof of Theorem 4.5 explicitly extends that prior framework. Section 7 makes the substantive role of the new loss explicit: the old objective (28) does not have V̂ as its minimizer in the input-output setting, so the introduction of (4) is a genuine modification rather than a renaming or an ansatz smuggled in by citation. No fitted constant is relabeled as a prediction; the convergence statements are asymptotic in network size and signature truncation, with no data-dependent fitted quantity that forces d_k(V̂, G) = 0 by construction. The experiments are validated against independent references such as particle filters, the Kalman filter, and analytically computed conditional expectations. The score is set to 2 rather than 0 only because the paper leans heavily on the authors' earlier NJODE results; that reliance is real evidence, not circular support. A separate correctness risk (not a circularity) is whether Example 5.5's threshold-classification function satisfies the joint continuity required for the uniform approximation argument at t = τ(t) when W_{τ(t)} = α; this concerns the hypotheses of the proof, not a circular derivation.

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

The theoretical claim introduces no fitted constants. The proof relies on standard approximation theorems and on the stated smoothness and independence assumptions. Experimental hyperparameters are not part of the central claim.

assumptions (6)
  • domain assumption Assumption 2: the output process V is almost surely not observed at a jump time
    Used to identify V_{t_i-} with V_{t_i} in the loss and to guarantee Vhat is the unique minimizer (Section 2.3, used in Theorem 4.4 Step 1).
  • domain assumption Assumption 3: the conditional expectation functions F_j are continuous and differentiable in t with integrable derivatives f_j
    Required for the signature and neural approximation argument in Theorem 4.4 Step 2; restricts the class of filtering problems.
  • domain assumption Assumption 6: the process Z is independent of the observation times and mask (n, t_i, M_i)
    Allows the Doob-Dynkin representation and the decomposition in Lemma 4.2; can be relaxed to conditional independence as noted in Remark 2.3.
  • standard math Universal approximation theorem for feed-forward neural networks (Hornik 1989, 1991)
    Used to approximate F_j and the compensation function in L2(mu) on compact sets in Theorem 4.4 Step 2.
  • standard math Signature approximation theorem for continuous functions on compact path sets with finite-dimensional augmentation (Krach et al. 2022, Prop. 3.8)
    Reduces path-dependent F_j to a function of truncated signature; cited from prior work.
  • standard math Compactness of the set A_N of piecewise linear bounded paths (Bugajewski and Gulgowski 2020)
    Used to define the compact path set P_epsilon in the approximation step.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Nonparametric Filtering, Estimation and Classification using Neural Jump ODEs." pith.science (2026). https://pith.science/paper/24SMSRWH

@misc{pith2026241203271,
  author       = {Pith},
  title        = {Pith review of: Nonparametric Filtering, Estimation and Classification using Neural Jump ODEs},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/24SMSRWH}},
  note         = {Machine review of arXiv:2412.03271}
}
abstract

Neural Jump ODEs model the conditional expectation between observations by neural ODEs and jump at arrival of new observations. They have demonstrated effectiveness for fully data-driven online forecasting in settings with irregular and partial observations, operating under weak regularity assumptions. This work extends the framework to input-output systems, enabling direct applications in online filtering and classification. We establish theoretical convergence guarantees for this approach, providing a robust solution to $L^2$-optimal filtering. Empirical experiments highlight the model's superior performance over classical parametric methods, particularly in scenarios with complex underlying distributions. These results emphasise the approach's potential in time-sensitive domains such as finance and health monitoring, where real-time accuracy is crucial.

Figures

Figures reproduced from arXiv: 2412.03271 by the authors.

Figure 1
Figure 1. Predicted and true conditional expectation [PITH_FULL_IMAGE:figures/full_fig_p023_1.png] view at source ↗
Figure 2
Figure 2. Predicted conditional expectation of the uncertain parameters of a geometric Brownian motion on [PITH_FULL_IMAGE:figures/full_fig_p024_2.png] view at source ↗
Figure 3
Figure 3. Distribution of prediction error of our model and reference methods at evaluation times [PITH_FULL_IMAGE:figures/full_fig_p024_3.png] view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: Test samples of parameter predictions of the CIR processes in Experiment 1 (top left), Experiment [PITH_FULL_IMAGE:figures/full_fig_p028_4.png]
Figure 5
Figure 5. Figure 5: Predicted conditional expectation of the signal process [PITH_FULL_IMAGE:figures/full_fig_p028_5.png]
Figure 6
Figure 6. Figure 6: Predicted and true conditional probability of the input process [PITH_FULL_IMAGE:figures/full_fig_p029_6.png]
Figure 7
Figure 7. Figure 7: The jump loss Ψjump converges faster and to a smaller value when trained with the original loss function (with L 1 aspect) (28) than with the pure L 2 IO loss function (4). 8 Conclusion In this work, we extended the framework of Neural Jump ODEs to input-output systems…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

48 extracted references · 35 canonical work pages

  1. [1]

    Extending path-dependent NJ - ODE s to noisy observations and a dependent observation framework

    William Andersson, Jakob Heiss, Florian Krach, and Josef Teichmann. Extending path-dependent NJ - ODE s to noisy observations and a dependent observation framework. Transactions on Machine Learning Research, 2024. ISSN 2835-8856. URL https://openreview.net/forum?id=0T2OTVCCC1

  2. [2]

    Black box variational inference for state space models

    Evan Archer, Il Memming Park, Lars Buesing, John Cunningham, and Liam Paninski. Black box variational inference for state space models. arXiv preprint arXiv:1511.07367, 2015

  3. [3]

    Clear: Calibrated learning for epistemic and aleatoric risk, 2025

    Ilia Azizi, Juraj Bodik, Jakob Heiss, and Bin Yu. Clear: Calibrated learning for epistemic and aleatoric risk, 2025. URL https://arxiv.org/abs/2507.08150

  4. [4]

    On the characterization of compactness in the space of functions of bounded variation in the sense of Jordan

    Dariusz Bugajewski and Jacek Gulgowski. On the characterization of compactness in the space of functions of bounded variation in the sense of Jordan . Journal of Mathematical Analysis and Applications, 484 0 (2), 2020

  5. [5]

    Multitask learning

    Rich Caruana, Lorien Pratt, and Sebastian Thrun. Multitask learning. Machine Learning 1997 28:1, 28: 0 41--75, 1997. ISSN 1573-0565. doi:10.1023/A:1007379606734. URL https://link.springer.com/article/10.1023/A:1007379606734

  6. [6]

    A primer on the signature method in machine learning

    Ilya Chevyrev and Andrey Kormilitzin. A primer on the signature method in machine learning. arXiv, 2016

  7. [7]

    Stochastic calculus and applications

    Samuel N Cohen and Robert James Elliott. Stochastic calculus and applications. Springer, 2015

  8. [8]

    Differentiable particle filtering via entropy-regularized optimal transport

    Adrien Corenflos, James Thornton, George Deligiannidis, and Arnaud Doucet. Differentiable particle filtering via entropy-regularized optimal transport. In International Conference on Machine Learning, pp.\ 2100--2111. PMLR, 2021

Show all 48 references
  1. [9]

    A theory of the term structure of interest rates

    John C Cox, Jonathan E Ingersoll, and Stephen A Ross. A theory of the term structure of interest rates. Econometrica, 53 0 (2): 0 385--407, 1985. ISSN 00129682, 14680262

  2. [10]

    On the strong law of large numbers for pairwise independent random variables

    S \'a ndor Cs \"o rg o , K \'a roly Tandori, and Vilmos Totik. On the strong law of large numbers for pairwise independent random variables. Acta Mathematica Hungarica, 42: 0 319--330, 1983

  3. [11]

    Universal approximation theorems for continuous functions of càdlàg paths and Lévy -type signature models

    Christa Cuchiero, Francesca Primavera, and Sara Svaluto-Ferro. Universal approximation theorems for continuous functions of càdlàg paths and Lévy -type signature models. Finance and Stochastics, 29 0 (2): 0 289--342, April 2025. ISSN 1432-1122. doi:10.1007/s00780-025-00557-5

  4. [12]

    Particle filtering

    Petar M Djuric, Jayesh H Kotecha, Jianqui Zhang, Yufei Huang, Tadesse Ghirmai, M \'o nica F Bugallo, and Joaquin Miguez. Particle filtering. IEEE signal processing magazine, 20 0 (5): 0 19--38, 2003

  5. [13]

    Term structures of credit spreads with incomplete accounting information

    Darrell Duffie and David Lando. Term structures of credit spreads with incomplete accounting information. Econometrica, 69 0 (3): 0 633--664, 2001

  6. [14]

    Multivariate statistics: A vector space approach

    Morris L Eaton. Multivariate statistics: A vector space approach. Institute of Mathematical Statistics, 2007

  7. [15]

    Embedding and learning with signatures

    Adeline Fermanian. Embedding and learning with signatures. Computational Statistics & Data Analysis, 157: 0 107148, 2021

  8. [16]

    General dynamic term structures under default risk

    Claudio Fontana and Thorsten Schmidt. General dynamic term structures under default risk. Stochastic Processes and their Applications, 128 0 (10): 0 3353--3386, 2018

  9. [17]

    Pricing corporate securities under noisy asset information

    R \"u diger Frey and Thorsten Schmidt. Pricing corporate securities under noisy asset information. Mathematical Finance, 19 0 (3): 0 403--421, 2009

  10. [18]

    Pricing and hedging of credit derivatives via the innovations approach to nonlinear filtering

    R \"u diger Frey and Thorsten Schmidt. Pricing and hedging of credit derivatives via the innovations approach to nonlinear filtering. Finance and Stochastics, 16 0 (1): 0 105--133, 2012

  11. [19]

    Dynamic defaultable term structure modeling beyond the intensity paradigm

    Frank Gehmlich and Thorsten Schmidt. Dynamic defaultable term structure modeling beyond the intensity paradigm. Mathematical Finance, 28 0 (1): 0 211--239, 2018

  12. [20]

    Deep dynamic modeling with just two time points: Can we still allow for individual trajectories? Biometrical Journal, 64 0 (8): 0 1426--1445, 2022

    Maren Hackenberg, Philipp Harms, Michelle Pfaffenlehner, Astrid Pechmann, Janbernd Kirschner, Thorsten Schmidt, and Harald Binder. Deep dynamic modeling with just two time points: Can we still allow for individual trajectories? Biometrical Journal, 64 0 (8): 0 1426--1445, 2022

  13. [21]

    Investigating a domain adaptation approach for integrating different measurement instruments in a longitudinal clinical registry

    Maren Hackenberg, Michelle Pfaffenlehner, Max Behrens, Astrid Pechmann, Janbernd Kirschner, and Harald Binder. Investigating a domain adaptation approach for integrating different measurement instruments in a longitudinal clinical registry. Biometrical Journal, 67 0 (1): 0 e70...

  14. [22]

    Inductive Bias of Neural Networks and Selected Applications

    Jakob Heiss. Inductive Bias of Neural Networks and Selected Applications. Doctoral thesis, ETH Zurich, Zurich, 2024. URL https://www.research-collection.ethz.ch/handle/20.500.11850/699241

  15. [23]

    How infinitely wide neural networks can benefit from multi-task learning - an exact macroscopic characterization

    Jakob Heiss, Josef Teichmann, and Hanna Wutte. How infinitely wide neural networks can benefit from multi-task learning - an exact macroscopic characterization. arXiv preprint arXiv:2112.15577, 2022. doi:10.3929/ETHZ-B-000550890. URL http://hdl.handle.net/20.500.11850/550890

  16. [24]

    Neural jump ordinary differential equations: Consistent continuous-time prediction and filtering

    Calypso Herrera, Florian Krach, and Josef Teichmann. Neural jump ordinary differential equations: Consistent continuous-time prediction and filtering. In International Conference on Learning Representations, 2021. URL https://openreview.net/forum?id=JFKR3WqwyXR

  17. [25]

    Approximation capabilities of multilayer feedforward networks

    Kurt Hornik. Approximation capabilities of multilayer feedforward networks. Neural networks, 4 0 (2): 0 251--257, 1991

  18. [26]

    Multilayer feedforward networks are universal approximators

    Kurt Hornik, Maxwell Stinchcombe, and Halbert White. Multilayer feedforward networks are universal approximators. Neural Networks, 2, 1989

  19. [27]

    Serial order: A parallel distributed processing approach

    Michael I Jordan. Serial order: A parallel distributed processing approach. In Advances in Psychology, volume 121. Elsevier, 1997

  20. [28]

    Foundations of modern probability

    Olav Kallenberg. Foundations of modern probability. Springer, 3rd edition, 2021

  21. [29]

    A new approach to linear filtering and prediction problems [j]

    Rudolph E Kalman et al. A new approach to linear filtering and prediction problems [j]. Journal of basic Engineering, 82 0 (1): 0 35--45, 1960

  22. [30]

    Introduction to Stochastic Calculus

    Rajeeva L Karandikar and Bhamidi V Rao. Introduction to Stochastic Calculus. Indian Statistical Institute Series. Springer Singapore, Singapore, 2018. ISBN 978-981-10-8317-4. doi:10.1007/978-981-10-8318-1. URL http://link.springer.com/10.1007/978-981-10-8318-1

  23. [31]

    Deep variational bayes filters: Unsupervised learning of state space models from raw data

    Maximilian Karl, Maximilian Soelch, Justin Bayer, and Patrick Van der Smagt. Deep variational bayes filters: Unsupervised learning of state space models from raw data. arXiv preprint arXiv:1605.06432, 2016

  24. [32]

    Estimation of the parameters of the stochastic differential equations black-scholes model share price of gold

    Khaldi Khaled and Meddahi Samia. Estimation of the parameters of the stochastic differential equations black-scholes model share price of gold. Journal of Mathematics and Statistics, 6 0 (4): 0 421, 2010

  25. [33]

    Kernels for sequentially ordered data

    Franz J Kiraly and Harald Oberhauser. Kernels for sequentially ordered data. Journal of Machine Learning Research, 20 0 (31): 0 1--45, 2019

  26. [34]

    Neural Jump Ordinary Differential Equations

    Florian Krach. Neural Jump Ordinary Differential Equations. Doctoral thesis, ETH Zurich, Zurich, 2025

  27. [35]

    Learning chaotic systems and long-term predictions with neural jump odes

    Florian Krach and Josef Teichmann. Learning chaotic systems and long-term predictions with neural jump odes. arXiv preprint arXiv:2407.18808, 2024

  28. [36]

    Optimal estimation of generic dynamics by path-dependent neural jump ODEs

    Florian Krach, Marc N \"u bel, and Josef Teichmann. Optimal estimation of generic dynamics by path-dependent neural jump ODEs . arXiv preprint arXiv:2206.14284, 2022

  29. [37]

    Deep Kalman filters

    Rahul G Krishnan, Uri Shalit, and David Sontag. Deep Kalman filters. arXiv preprint arXiv:1511.05121, 2015

  30. [38]

    Structured inference networks for nonlinear state space models

    Rahul G Krishnan, Uri Shalit, and David Sontag. Structured inference networks for nonlinear state space models. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 31, 2017

  31. [39]

    Variational marginal particle filters

    Jinlin Lai, Justin Domke, and Daniel Sheldon. Variational marginal particle filters. In International Conference on Artificial Intelligence and Statistics, pp.\ 875--895. PMLR, 2022

  32. [40]

    Auto-encoding sequential monte carlo

    Tuan Anh Le, Maximilian Igl, Tom Rainforth, Tom Jin, and Frank Wood. Auto-encoding sequential monte carlo. arXiv, 2017

  33. [41]

    Filtering variational objectives

    Chris J Maddison, John Lawson, George Tucker, Nicolas Heess, Mohammad Norouzi, Andriy Mnih, Arnaud Doucet, and Yee Teh. Filtering variational objectives. Advances in Neural Information Processing Systems, 30, 2017

  34. [42]

    On the pricing of corporate debt: The risk structure of interest rates

    Robert C Merton. On the pricing of corporate debt: The risk structure of interest rates. The Journal of finance, 29 0 (2): 0 449--470, 1974

  35. [43]

    Variational sequential monte carlo

    Christian Naesseth, Scott Linderman, Rajesh Ranganath, and David Blei. Variational sequential monte carlo. In International conference on artificial intelligence and statistics, pp.\ 968--977. PMLR, 2018

  36. [44]

    Kalman filtering: Processing all measurements together vs processing them sequentially

    Ralff. Kalman filtering: Processing all measurements together vs processing them sequentially. Mathematics Stack Exchange, 2021. URL https://math.stackexchange.com/q/4058151. URL: https://math.stackexchange.com/q/4058151 (version: 2021-03-11)

  37. [45]

    Kalmannet: Neural network aided kalman filtering for partially known dynamics

    Guy Revach, Nir Shlezinger, Xiaoyong Ni, Adria Lopez Escoriza, Ruud JG Van Sloun, and Yonina C Eldar. Kalmannet: Neural network aided kalman filtering for partially known dynamics. IEEE Transactions on Signal Processing, 70: 0 1532--1547, 2022

  38. [46]

    Learning internal representations by error propagation

    David E Rumelhart, Geoffrey E Hinton, and Ronald J Williams. Learning internal representations by error propagation. Technical report, California Univ San Diego La Jolla Inst for Cognitive Science, 1985

  39. [47]

    A structural model with unobserved default boundary

    Thorsten Schmidt and Alexander Novikov. A structural model with unobserved default boundary. Applied mathematical finance, 15 0 (2): 0 183--203, 2008

  40. [48]

    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 gl...

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.