Pith. sign in

REVIEW 5 major objections 6 minor 36 references

History Matching under Uncertainty of Geological Scenarios with Implicit Geological Realism Control with Generative Deep Learning and Graph Convolutions

T0 review · 5 major / 6 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read A graph-based Wasserstein autoencoder with a pull-back metric realism loss can history-match reservoir models across two geological scenarios while keeping outputs geologically plausible.

desk verdict A plausible, clearly-presented graph-VAE approach to multi-scenario history matching whose realism-control term needs quantitative validation before the central claim is accepted. read the letter →

arxiv 2507.10201 v1 pith:WWNURBVQ submitted 2025-07-14 stat.AP cs.LGphysics.data-an

Gleb Shishaev , Vasily Demyanov , Daniel Arnold This is my paper · ORCID
classification stat.APcs.LGphysics.data-an
keywords graphconvolutionalnetworkWassersteinautoencoderlatentspacegeologicalrealismhistorymatchinguncertaintyquantificationgeodesicmetricreservoirmodelling
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

History matching usually searches over model parameters and can drift into geologically implausible territory once it starts matching production data. This paper argues that a graph-based Wasserstein autoencoder, which compresses thousands of porosity and permeability values into a 30-dimensional latent space, can keep the search realistic by adding a realism term to the objective. The realism term is the decoder's pull-back metric $Z_f(z)=J_\mu^\top J_\mu+J_\sigma^\top J_\sigma$, a curvature-like quantity that is small where training realisations cluster and large in empty regions, so penalising it discourages the optimiser from leaving geologically plausible zones. Using a synthetic dataset of channelised models with one- and two-channel scenarios, the authors show that optimisation through this latent space matches static and dynamic well data, and an ablation study shows that switching off the realism term lets the optimiser land on unrealistic models. The wider point is that geological realism can be imposed implicitly through the geometry of a learned latent space rather than through hand-coded geological constraints.

What carries the argument

The machine that carries the argument is the graph-based Wasserstein autoencoder together with its realism-aware objective. The autoencoder uses hierarchical graph convolutions (with local and global neighbourhoods of node subsets) rather than lattice convolutions, so it can represent the topology of curvilinear channels directly. The latent space is treated as a Riemannian manifold, and the decoder's pull-back metric $Z_f(z)=J_f(z)^\top J_f(z)=J_\mu(z)^\top J_\mu(z)+J_\sigma(z)^\top J_\sigma(z)$ is used as a density proxy: small in regions where training data cluster, large where no training data exist. The history-matching objective is $Loss = w_f Loss_{\text{flow}} + w_s Loss_{\text{static}} + w_r Loss_{\text{realism}}$, with the realism term evaluated as $Z_f(z)$ at the current latent vector; this is what steers CMA-ES toward the geologically plausible part of the space. Geodesic distance under this metric, rather than Euclidean distance, is used for interpolation tests.

What would settle it

Compute $Z_f(z)$ on a grid of latent points and compare it with a kernel-density estimate of the encoder's distribution over the training set; if high metric values are not concentrated where training samples are sparse, the realism loss is not measuring what the paper claims. A second check is to run the ablation with the realism term disabled on several reference models; if the optimiser still stays inside the high-density cloud, the realism signal is redundant rather than responsible for staying realistic.

Watch

Extended reading notes

Core claim

The paper claims that one shared latent space, learned by a graph-based Wasserstein autoencoder, can simultaneously represent multiple geological scenarios and support history matching with implicit realism control. The encoder maps each 3D porosity-permeability model (3840 values) to a 30-dimensional latent vector; the decoder acts as the forward model. The objective function combines the mismatch of dynamic production data, mismatch of static well-log data, and a realism loss computed as the decoder's pull-back metric $Z_f(z)$; the metric is meant to estimate latent-space density, so low values keep the optimiser near the training cloud of geologically realistic realisations. Experiments on 5000 synthetic channelised models with single- and double-channel scenarios show that CMA-ES optimisation in this latent space finds realisations close to the reference model's static and dynamic well responses, while geodesic interpolation produces more realistic intermediate models than Euclidean interpolation. The ablation study, with the realism term disabled, shows the optimiser drifting away from the prior cloud and producing a geologically unrealistic model despite matching the data.

Load-bearing premise

The realism control rests on the assumption that the decoder's pull-back metric $Z_f(z)$ measures latent-space density, small near training realisations and large in empty regions; the paper adopts this from prior work (Section 2.3) without validating it on its synthetic dataset, and it also assumes the training ensemble contains only geologically realistic models.

Editorial extensions

If this is right

  • If the GWAE claims hold, history matching no longer requires a separate geological-constraint tool: the same decoder used for generation supplies the realism signal through its Jacobian.
  • Because all 2500 single-channel and 2500 double-channel realisations share one 30-dimensional latent space, scenario uncertainty (not just parameter uncertainty inside one scenario) can be explored and conditioned in a single optimisation run.
  • Geodesic interpolation staying in dense regions means the latent space can serve as a principled way to move between geological concepts, with intermediate models remaining mostly realistic.
  • Soft conditioning to static and dynamic well data replaces hard geostatistical constraints, so imperfect or uncertain well-log data can be incorporated during the search.
  • The 3840-to-30 dimensionality reduction makes the inverse problem tractable for population-based optimisers such as CMA-ES.

Reading between the lines

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

  • If the metric-to-density correspondence holds beyond this synthetic case, the same realism loss could be applied to real curvilinear grid geometries, where the graph representation is most useful.
  • A direct test of the mechanism would replace the pull-back metric with an explicit latent-density estimate and compare the history-matching ensembles; that would isolate the realism control's source.
  • The shared latent space suggests a continuum of intermediate models between one-channel and two-channel scenarios; whether those intermediates correspond to meaningful geology would require field data with known sedimentology.
  • Comparing the graph-based autoencoder against a lattice-based VAE on the same dataset would quantify how much of the improvement comes from graph convolutions rather than from the realism term.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

5 major / 6 minor

Summary. The paper proposes a graph-based Wasserstein autoencoder (GWAE) for history matching in reservoir models, using a latent space shared by two synthetic channelized geological scenarios. The methodology combines graph convolutions, a WAE objective, a pull-back metric for latent-space geometry, and a three-term objective for history matching. Experiments on a synthetic dataset of 5000 realizations illustrate generative quality, latent-space structure via PCA, t-SNE, and TDA, geodesic interpolation, and history matching with and without the realism term. The central claim is that the realism term implicitly controls geological realism by penalizing low-density latent regions.

Significance. If the realism mechanism were validated, the paper would offer a practical way to constrain generative-model history matching without explicit geological rules, and the graph representation is a plausible improvement over lattice-based CNNs for curvilinear features. Strengths include a complete synthetic experiment pipeline with flow simulation, an ablation study, and structural analysis of the latent space. However, the central evidence is largely qualitative, and the specific assumption connecting decoder Jacobians to latent density is load-bearing and currently unvalidated; therefore the significance is conditional on additional quantitative validation.

major comments (5)
  1. [Section 2.3/2.4, Eqs. (6)-(7)] The realism loss is the load-bearing component of the paper's central claim, but the asserted relationship between the decoder pull-back metric Zf(z)=Jmu^T Jmu + Jsigma^T Jsigma and the density of the latent training distribution is not validated on the presented dataset. In a Wasserstein autoencoder the encoder is regularized toward a Gaussian prior, so low Jacobian norm in high-density latent regions is not guaranteed; the Jacobian can be large near class boundaries inside the manifold and small far from data. The manuscript should add a quantitative check, for example comparing Zf(z) with a kernel density estimate of latent training samples or with reconstruction error of decoded samples, before Eq. (7) can support the realism-control claim. Additionally, Zf(z) is matrix-valued, and the text does not state which scalarization (trace, determinant, or spectral norm) enters Eq. (7), so the direction of the penalty is not uniquely defined.
  2. [Section 3.6, Figs. 11-13] The ablation demonstrates only that removing the realism term changes the PCA location and visual appearance of the optimized model; it does not quantify geological realism. The 'with realism' solutions are not scored with any realism metric independent of the loss itself (e.g., facies connectivity, channel geometry statistics, or distance to the training manifold), so the experiment cannot exclude the possibility that the realism term simply pulls solutions toward the latent mean. Please add quantitative realism metrics and repeat the optimization several times to report variability across runs.
  3. [Section 3.5, Figs. 8-10] The history-matching demonstration is assessed visually, with statements such as 'quite close' and 'acceptable'; no quantitative data-match metrics (e.g., normalized RMS error, ensemble coverage of the reference production profiles, or static well mismatch) are reported. Since the paper claims conditioning on static and dynamic well data, quantitative misfit values and associated uncertainty ranges are needed to support that claim.
  4. [Section 2.4] The claim that the realism loss measures geological realism depends on the assumption that the training ensemble consists only of geologically realistic models. The paper states this assumption but does not discuss its consequences: Zf can at best measure proximity to the synthetic prior, not geological plausibility in an absolute sense. Please state this limitation explicitly and, if possible, validate against an independent realism criterion.
  5. [Section 3.2] The latent dimensionality is set to 30 with the statement that lower values deteriorate performance, but no sensitivity analysis or quantitative evidence (e.g., reconstruction error versus latent dimension) is provided. Because the latent dimension controls the search space geometry and the realism metric, this choice should be justified with data.
minor comments (6)
  1. [Abstract and Section 1] The word 'scenarious' appears in the abstract and elsewhere; please use 'scenarios' consistently.
  2. [Section 2.4, Eq. (7)] The weights in Eq. (7) are listed as wf, wr, wr; the second and third should presumably be ws and wr.
  3. [Figures 4 and 6] The TDA plots are described qualitatively without axis labels or a statement of which persistence measures are shown; please add axis labels and a brief explanation of H0 and H1 for readers unfamiliar with persistence diagrams.
  4. [Figures 8 and 10] Please add color scales and units for the porosity maps and well plots to make the visual comparisons reproducible.
  5. [Section 3.4] The claim that geodesic interpolation stays in dense regions of the latent space is not quantified; reporting the integrated pull-back metric along the Euclidean and geodesic paths would make the comparison more concrete.
  6. [General] No code or data availability statement is included; providing the trained model and the synthetic dataset construction script would improve reproducibility.

Circularity Check

2 steps flagged · score 6.0 of 10

The 'geological realism' control is defined as a latent-density penalty (Eq. 6), so the ablation showing it keeps optimizers near the prior manifold is a check of the loss by construction, not an independent validation of geological realism.

  1. self definitional [Section 2.3, Eq. (6), and Section 2.4, 'Realism loss function']
    "By definition, the Decoder is a function of the form: f(Z) = P(X|Z) = µθ + σθ ⊙ ϵ ... Zf (z) = Jf (z)⊺Jf (z) = Jµ(z)⊺Jµ(z) + Jσ(z)⊺Jσ(z). (6) ... The values of the metric tensor should be small in regions of Z confirmed by observations x, otherwise Z should be large ... We replaced the probability estimation with an estimate of the density of LS Zf (z) from the equation 6. Thus, as the third term, the loss function will receive an estimate of density Zf (z)."

    The realism loss is not an independent measure of geological realism; it is defined to be a density estimate of the latent training distribution. Minimizing Lossrealism keeps the optimizer in regions where the prior training ensemble is dense by construction. The ablation result—that adding the realism term keeps solutions near the prior manifold and removing it lets them leave—is therefore an automatic consequence of the loss definition. The identification of 'high density' with 'geologically realistic' is imported from the assumption that the training set consists only of realistic scenarios (Section 2.4), so the realism-control test reduces to a consistency check of the loss against its own definition, not a validation of geological plausibility.

  2. other [Section 3.6, Ablation study]
    "The left part of Figure 11 shows the spatial location of the geological representation found in the PCA space. It can be seen that the found location (blue dot) is significantly distant from the main group of points (grey dots - geologically consistent prior manifold). This suggests that the reconstructed geological realisation will be significantly different from the prior set, which is a violation of geological realism."

    This ablation interprets distance from the prior manifold as a violation of geological realism. But the realism loss was explicitly designed to penalize exactly such distances, via its construction as an estimate of latent density. Thus the ablation demonstrates that the penalty moves points toward the training manifold, which is what the loss was designed to do; it does not independently establish that the pull-back metric Zf(z) is a valid proxy for geological realism. The reasoning is circular: the metric is assumed to track realism, and then behaviour consistent with that assumption is reported as evidence that it tracks realism.

full rationale

The paper's core novelty is the implicit geology-realism control through the pull-back metric Zf(z) in Eq. (6). The metric is asserted, following prior work [3], to be small in dense latent regions and large in empty ones; the realism loss is then set, by definition, to this metric (Section 2.4). Consequently, the demonstration that adding the realism term keeps optimised solutions inside the PCA projection of the training manifold, and that removing it lets the solution drift away (Section 3.6, Fig. 11), is expected from the loss construction: the loss is a density penalty, so it penalises low-density regions. The paper's further equation of 'dense prior regions' with 'geologically realistic models' rests on the unvalidated assumption that the training ensemble contains only geologically realistic realisations. This makes the realism-control evidence partially self-definitional rather than an independent test of geological plausibility. However, other parts of the paper are not circular: the graph-based Wasserstein autoencoder is an externally defined architecture with its own training objective; the history-matching results are assessed against a reference model using static and dynamic well data, which is an independent external check; and the latent-space analyses (PCA, t-SNE, TDA) are descriptive rather than load-bearing for the circularity. The self-citations in the bibliography are not used to prove the core mechanism, so they do not contribute to the circularity score. Overall, the central realism-control claim partially reduces to its own definition, warranting a score of 6, but the paper retains independent content in the graph architecture and the data-matching experiments.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

The central claim rests on the choice of latent dimensionality, the three loss weights, the assumption that the pull-back metric measures data density, and the assumption that the synthetic prior contains only realistic geology. The metric-based realism claim is partly self-referential because realism is evaluated by proximity to the same latent manifold the loss enforces.

free parameters (3)
  • Latent space dimensionality = 30
    Set by hand in Section 3.2; paper states that reducing below 30 degrades encoder performance, but no quantitative evidence is shown.
  • Objective loss weights wf, ws, wr = not reported
    Equation 7 defines the three-component loss, but the values of the weights are never given; they are degrees of freedom that affect the trade-off between data match and realism.
  • Training hyperparameters (layers, hidden units, learning rate, epochs) = not reported
    The architecture is described at a high level, but no training hyperparameters are provided, so the method is under-specified.
assumptions (4)
  • domain assumption The training ensemble consists only of geologically realistic scenarios.
    Stated in Section 2.4: "The GW AE training set is assumed to consist only of geologically realistic scenarios." This is not validated against external geological criteria; if false, the autoencoder learns an unrealistic prior.
  • domain assumption The pull-back metric Zf(z) = Jf^T Jf is small in data-dense regions and large in data-sparse regions of the latent space.
    Adapted from Arvanitidis (ref [3]) in Section 2.3; the paper assumes this holds for the trained GWAE decoder so that minimizing Zf controls realism. No empirical validation of the metric-density correspondence is given.
  • standard math The decoder output is of the form f(z)=mu_theta + sigma_theta * epsilon with epsilon ~ N(0,I), and the Jacobians exist.
    Assumed in Section 2.3 for the pull-back metric computation; standard in VAE-type models but the Jacobian approximation is an assumption.
  • domain assumption A 30-dimensional latent space is sufficient to represent both one-channel and two-channel scenarios and the petrophysical relationships.
    The paper chooses 30 dimensions based on informal degradation observations (Section 3.2); this is an assumption about the intrinsic dimensionality of the geological prior.

how reviews work

0 comments
Cite this review

Pith. "Pith review of History Matching under Uncertainty of Geological Scenarios with Implicit Geological Realism Control with Generative Deep Learning and Graph Convolutions." pith.science (2026). https://pith.science/paper/WWNURBVQ

@misc{pith2026250710201,
  author       = {Pith},
  title        = {Pith review of: History Matching under Uncertainty of Geological Scenarios with Implicit Geological Realism Control with Generative Deep Learning and Graph Convolutions},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/WWNURBVQ}},
  note         = {Machine review of arXiv:2507.10201}
}
read the original abstract

The graph-based variational autoencoder represents an architecture that can handle the uncertainty of different geological scenarios, such as depositional or structural, through the concept of a lowerdimensional latent space. The main difference from recent studies is utilisation of a graph-based approach in reservoir modelling instead of the more traditional lattice-based deep learning methods. We provide a solution to implicitly control the geological realism through the latent variables of a generative model and Geodesic metrics. Our experiments of AHM with synthetic dataset that consists of 3D realisations of channelised geological representations with two distinct scenarios with one and two channels shows the viability of the approach. We offer in-depth analysis of the latent space using tools such as PCA, t-SNE, and TDA to illustrate its structure.

Figures

Figures reproduced from arXiv: 2507.10201 by the authors.

Figure 1
Figure 1. A latent vector is sampled from the LS, which is decoded to the geological model. Three-component OF then calculated and provides information for the opti￾miser about the next optimisation step in the LS This approach of HM through a LS provides the following opportunities: – Provide common parameterisation across different scenarious, while also cap￾turing complex, non-linear relationships between input parameters;… view at source ↗
Figure 2
Figure 2. Settings of the channelised synthetic dataset. Two distinct geological scenarios: single and double channels are considered. The second row represents some examples from the dataset Five thousand realisations were created, 2500 for each scenario with one or two channels. The dimension of the model is x = 16, y = 12, z = 10, which means that the dimension of the optimisation search space is x · y · z · [PITH_FULL_IM… view at source ↗
Figure 3
Figure 3. 1: Examples of the initial porosity grid from the training dataset. 2: GWAE reconstructed. 3: GWAE preserves petrophysical relations and spatial propagation of properties under uncertainty of geological scenarios [PITH_FULL_IMAGE:figures/full_fig_p012_3.png] view at source ↗
Figures from the paper (10 more)
Figure 4
Figure 4. Figure 4: Visualisation of the 30-dimensional LS in the PCA,t-SNE, and TDA spaces. Each point in the PCA and t-SNE space represents a 30-dimensional vector character￾ising a geological realisation from the training dataset. Blue dots depict a geological scenario with one channel…
Figure 5
Figure 5. Figure 5: 3-d t-SNE visualisation of the 30-d LS. Every point in this cloud is a sample from the training dataset. left: the LS in the scenario dimension, where red bars represent a scenario with one channel, and green bars represent two channels [PITH_FULL_IMAGE:figures/full_f…
Figure 6
Figure 6. Figure 6: The LS in the context of TDA. Red circle denote points that are furthest from the rest on H0, and green circles most stable rings (H1). However, it cannot be concluded that these objects are global topological properties of the space. TDA Analysing the topology of the …
Figure 7
Figure 7. Figure 7: Variation of ensemble members along the shortest path trajectories in the LS. Left side: starting and final points in the LS. Right side: Euclidean (E) and Geodesic (G) interpolation options in 10 steps. It’s noticeable that Geodesic interpolation enables the reconstru…
Figure 8
Figure 8. Figure 8: Visualisation of the reference porosity property (red dot) and the cubes ob￾tained through the optimisation process (blue dots). Due to high uncertainty in regions without well data, the geological realisations have not been completely similar to our reference model On…
Figure 9
Figure 9. Figure 9: Production dynamics of injectors and producers for the initial model ensemble (top row) in comparison to the final optimised ensemble (bottom row) vs the reference (true) dynamic response Earlier, we considered the static well properties known to us, so the optimisa￾ti…
Figure 10
Figure 10. Figure 10: Static (porosity) property in wells P2, I2, I5: a. examples of initial (start) realisations, b. reference (true) realisation, c. optimised (final) realisations Loss component of the OF, which implicitly measures the realism of GWAE generated realisation, was disabled,…
Figure 11
Figure 11. Figure 11: Left: spatial location of the geological model (blue dot) in the PCA represen￾tation. Right: the reconstructed geological model (blue dot) differs from the reference (red dot) and models with Realism Loss component (grey dots) [PITH_FULL_IMAGE:figures/full_fig_p018_11.png]
Figure 12
Figure 12. Figure 12: Static (porosity) property in wells P2, I2, I5. Reference model (red dot), no- (blue dot) and with (grey dots) Realism Loss component. The overall similarity of well logs can be seen [PITH_FULL_IMAGE:figures/full_fig_p019_12.png]
Figure 13
Figure 13. Figure 13: Production dynamics of injectors (I2, I5) and producer (P2) for the reference, no-Realism and with Realism Loss 4 Conclusions We have shown the application of graph variational autoencoders to the problem of HM with geological realism control. It is demonstrated that …

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

36 extracted references · 34 canonical work pages

  1. [1]

    Recent developments combining ensemble smoother and deep gen- erative networks for facies history matching

    Smith Arauco, Jose Castro, J´ ulia Potratz, Alexandre Emerick, and Marco Pacheco. Recent developments combining ensemble smoother and deep gen- erative networks for facies history matching. Computational Geosciences, On-line First, 02 2021

    work page 2021

  2. [2]

    Integration of en- semble data assimilation and deep learning for history matching facies mod- els

    Smith Arauco, Alexandre Emerick, and Marco Pacheco. Integration of en- semble data assimilation and deep learning for history matching facies mod- els. 10 2017

    work page 2017

  3. [3]

    Geometrical Aspects of Manifold Learning

    Georgios Arvanitidis. Geometrical Aspects of Manifold Learning. PhD the- sis, 2019

    work page 2019

  4. [4]

    Latent space oddity: on the curvature of deep generative models, 2021

    Georgios Arvanitidis, Lars Kai Hansen, and Søren Hauberg. Latent space oddity: on the curvature of deep generative models, 2021

    work page 2021

  5. [5]

    Variational autoencoder or generative adversarial networks? a comparison of two deep learning methods for flow and transport data assimilation

    Jichao Bao, Liangping Li, and Arden Davis. Variational autoencoder or generative adversarial networks? a comparison of two deep learning methods for flow and transport data assimilation. Mathematical Geosciences, 54, 05 2022

    work page 2022

  6. [6]

    Topology and data

    Gunnar Carlsson. Topology and data. Bulletin of The American Mathe- matical Society - BULL AMER MATH SOC , 46:255–308, 04 2009

    work page 2009

  7. [7]

    Elsheikh

    Shing Chan and Ahmed H. Elsheikh. Exemplar-based synthesis of geology using kernel discrepancies and generative neural networks, 2018

    work page 2018

  8. [8]

    Elsheikh

    Shing Chan and Ahmed H. Elsheikh. Parametric generation of conditional geological realizations using generative neural networks. Computational Geosciences, 23(5):925–952, jul 2019

    work page 2019

Show all 36 references
  1. [9]

    Elsheikh

    Shing Chan and Ahmed H. Elsheikh. Parametrization and generation of geological models with generative adversarial networks, 2019

    2019

  2. [10]

    Automated discovery of fundamental variables hidden in experimental data

    Boyuan Chen, Kuang Huang, Sunand Raghupathi, Ishaan Chandratreya, Qiang Du, and Hod Lipson. Automated discovery of fundamental variables hidden in experimental data. Nature Computational Science, 2:433–442, 07 2022

    2022

  3. [11]

    Uncertainty quantification in reservoir prediction: Part 2—handling uncer- tainty in the geological scenario

    Vasily Demyanov, Daniel Arnold, Temistocles Rojas, and Michael Christie. Uncertainty quantification in reservoir prediction: Part 2—handling uncer- tainty in the geological scenario. Mathematical Geosciences, 51, 07 2018

    2018

  4. [12]

    Generating realistic geology conditioned on physical measurements with generative adversarial networks, 2018

    Emilien Dupont, Tuanfeng Zhang, Peter Tilke, Lin Liang, and William Bai- ley. Generating realistic geology conditioned on physical measurements with generative adversarial networks, 2018

    2018

  5. [13]

    Edelsbrunner, D

    H. Edelsbrunner, D. Letscher, and A. Zomorodian. Topological persistence and simplification. In Proceedings 41st Annual Symposium on Foundations of Computer Science , pages 454–463, 2000

    2000

  6. [14]

    Guss and Ruslan Salakhutdinov

    William H. Guss and Ruslan Salakhutdinov. On characterizing the capacity of neural networks using algebraic topology, 2018

    2018

  7. [15]

    The cma evolution strategy: A tutorial, 2023

    Nikolaus Hansen. The cma evolution strategy: A tutorial, 2023

    2023

  8. [16]

    Analysis of a complex of statistical variables into principal components

    Harold Hotelling. Analysis of a complex of statistical variables into principal components. Journal of Educational Psychology , 24:498–520, 1933. 22 Gleb Shishaev, Vasily Demyanov, and Daniel Arnold

    1933

  9. [17]

    Deep convolutional autoencoders for robust flow model calibration under uncertainty in geologic continuity

    Anyue Jiang and Behnam Jafarpour. Deep convolutional autoencoders for robust flow model calibration under uncertainty in geologic continuity. Wa- ter Resources Research, 57, 10 2021

    2021

  10. [18]

    Training image-based scenario modeling of fractured reservoirs for flow uncertainty quantification

    Andre Jung, Darryl Fenwick, and Jef Caers. Training image-based scenario modeling of fractured reservoirs for flow uncertainty quantification. Com- putational Geosciences, 17, 12 2013

    2013

  11. [19]

    Introduction to pullback metric

    Andrii Khlevniuk and Vitalii Tymchyshyn. Introduction to pullback metric. 10 2017

    2017

  12. [20]

    Training- image based geostatistical inversion using a spatial generative adversarial neural network

    Eric Laloy, Romain H´ erault, Diederik Jacques, and Niklas Linde. Training- image based geostatistical inversion using a spatial generative adversarial neural network. Water Resources Research, 54(1):381–406, 2018

    2018

  13. [21]

    Inversion using a new low-dimensional representation of complex binary geological media based on a deep neural network

    Eric Laloy, Romain H´ erault, John Lee, Diederik Jacques, and Niklas Linde. Inversion using a new low-dimensional representation of complex binary geological media based on a deep neural network. Advances in Water Re- sources, 110:387–405, 2017

    2017

  14. [22]

    Lawrence

    Neil D. Lawrence. Probabilistic non-linear principal component analysis with gaussian process latent variable models. J. Mach. Learn. Res., 6:1783– 1816, 2005

    2005

  15. [23]

    Feature extraction using a deep learning algorithm for uncertainty quantification of channelized reservoirs

    Kyungbook Lee, Jungtek Lim, Seongin Ahn, and Jaejun Kim. Feature extraction using a deep learning algorithm for uncertainty quantification of channelized reservoirs. Journal of Petroleum Science and Engineering , 171:1007–1022, 2018

    2018

  16. [24]

    Maximum likelihood estimation of in- trinsic dimension

    Elizaveta Levina and Peter Bickel. Maximum likelihood estimation of in- trinsic dimension. In L. Saul, Y. Weiss, and L. Bottou, editors, Advances in Neural Information Processing Systems , volume 17. MIT Press, 2004

    2004

  17. [25]

    Hamilton, Jan Eric Lenssen, Gaurav Rattan, and Martin Grohe

    Christopher Morris, Martin Ritzert, Matthias Fey, William L. Hamilton, Jan Eric Lenssen, Gaurav Rattan, and Martin Grohe. Weisfeiler and leman go neural: Higher-order graph neural networks, 2021

    2021

  18. [26]

    Lukas Mosser, Olivier Dubrule, and Martin J. Blunt. Deepflow: History matching in the space of deep generative models, 2019

    2019

  19. [27]

    Karl Pearson. Liii. on lines and planes of closest fit to systems of points in space. The London, Edinburgh, and Dublin Philosophical Magazine and Journal of Science , 2(11):559–572, 1901

    1901

  20. [28]

    Rubenstein, Bernhard Schoelkopf, and Ilya Tolstikhin

    Paul K. Rubenstein, Bernhard Schoelkopf, and Ilya Tolstikhin. On the latent space of wasserstein auto-encoders, 2018

    2018

  21. [29]

    Thomas Fletcher

    Hang Shao, Abhishek Kumar, and P. Thomas Fletcher. The riemannian geometry of deep generative models, 2017

    2017

  22. [30]

    Emerick, and Marco Aur´ elio C

    Canchumuni Smith, Alexandre A. Emerick, and Marco Aur´ elio C. Pacheco. Towards a robust parameterization for conditioning facies models using deep variational autoencoders and ensemble smoother. Computers and Geo- sciences, 128:87–102, 2019

    2019

  23. [31]

    Geological realism in fluvial facies modelling with gan under variable depositional conditions

    Chao Sun, Vasily Demyanov, and Daniel Arnold. Geological realism in fluvial facies modelling with gan under variable depositional conditions. Computational Geosciences, 27, 01 2023

    2023

  24. [32]

    J. Tierny. Topological Data Analysis for Scientific Visualization . Mathe- matics and Visualization. Springer International Publishing, 2018

    2018

  25. [33]

    Wasserstein auto-encoders

    Ilya Tolstikhin, Olivier Bousquet, Sylvain Gelly, and Bernhard Schoelkopf. Wasserstein auto-encoders. 2017

    2017

  26. [34]

    Lawrence

    Alessandra Tosi, Søren Hauberg, Alfredo Vellido, and Neil D. Lawrence. Metrics for probabilistic geometries, 2014

    2014

  27. [35]

    Visualizing data using t-sne

    Laurens van der Maaten and Geoffrey Hinton. Visualizing data using t-sne. Journal of Machine Learning Research , 9(86):2579–2605, 2008

    2008

  28. [36]

    The choice of the loss function for solving the problem of history matching of the geological and simulation model: thesis for the degree of candidate of technical sciences: spec

    Grachik Yeremian. The choice of the loss function for solving the problem of history matching of the geological and simulation model: thesis for the degree of candidate of technical sciences: spec. 2.3.1 . PhD thesis, 2022

    2022

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.