Pith. sign in

REVIEW 3 major objections 5 minor 1 cited by

Variational Encoder-Decoders for Learning Latent Representations of Physical Systems

T0 review · 3 major / 5 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read A variational encoder-decoder can represent the map from 1,475 groundwater transmissivity parameters to 323 pressure observations with as few as 50 latent variables, below the 147 latent dimensions a linear CCA analysis suggests are…

desk verdict Applied VAE-style surrogate paper with a real dataset and sensible architecture, but the conditional-modeling claim is undone by an x-independent prior and the headline r=50 result is selected on the test set. read the letter →

arxiv 2412.05175 v1 pith:STHCZA7N submitted 2024-12-06 cs.LG stat.ML

Subashree Venkatasubramanian , David A. Barajas-Solano This is my paper · ORCID
classification cs.LGstat.ML
keywords variationalencoder-decoderlatentrepresentationdimensionreductiondisentanglementcovarianceregularizationgroundwaterflowsurrogatemodelingcanonicalcorrelationanalysis
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 aims to show that a variational encoder-decoder (VED) can learn a genuinely low-dimensional representation of a high-dimensional physical input-output map: specifically, the map from 1,475 log-transmissivity parameters of a groundwater flow model to 323 hydraulic pressure readings. In their experiments, a VED with only 50 latent variables reconstructs the pressure outputs almost as accurately as one with 100 or 200 variables, and it does so below the 147-dimension threshold that a linear canonical-correlation analysis needs to capture 95% of its explained variance. The authors further claim that small nonzero weights on the KL-divergence term and on a covariance-based disentanglement penalty improve the separation of latent features and let the decoder turn random Gaussian noise into realistic output distributions, without sacrificing reconstruction accuracy. If correct, this gives a recipe for building nonlinear surrogates and generative models of physics systems that are far more compact than linear dimension-reduction methods would allow.

What carries the argument

The central machinery is the encoder-decoder pair plus two regularizers. The encoder $q_\phi(z \mid x) = \mathcal{N}(g_\phi(x), \operatorname{diag}\{\exp h_\phi(x)\})$ compresses a 1,475-dimensional input to an $r$-dimensional code, and the decoder $p_\theta(y \mid z) = \mathcal{N}(f_\theta(z), \sigma^2 I_m)$ maps the code to the 323-dimensional output. Training maximizes the ELBO-like loss $\tfrac{1}{2}\text{MSE} + \beta\,\text{KLD} + \lambda\,\text{COV}(\phi)$, where $\beta = \sigma^2$ controls the KL weight and $\text{COV}(\phi)$ penalizes the squared distance of the aggregate encoding covariance $\operatorname{Cov}_{q_\phi(z)}[z]$ from the identity matrix (the DIP-VAE-II penalty), pushing the aggregate code distribution toward the prior $\mathcal{N}(0,I_r)$. The architectural workhorse is a deep residual-block convolutional encoder applied after an irregular-grid-to-Cartesian mapping, paired with a shallow two-layer decoder.

What would settle it

Retrain the same $r=50$ VED on the same training samples but with the output rows randomly permuted against the inputs, breaking the $x \to y$ correspondence; if the test MSE stays close to the unshuffled value, the encoder is not learning input-specific structure and the claim that the latent code compresses the $x \to y$ relationship fails. If instead the shuffled MSE rises substantially, the model is genuinely carrying input information through the code.

Watch

Extended reading notes

Core claim

Stated on the paper's own terms, the discovery is that a VED trained with the loss $L_{\beta,\lambda}(\theta,\phi) = \tfrac{1}{2}\text{MSE} + \beta\,\text{KLD} + \lambda\,\text{COV}(\phi)$ discovers a nonlinear low-dimensional code for the $x \to y$ relationship. On the Hanford groundwater benchmark, the VED reaches a test MSE around 0.05 at $r=50$, compared with about 0.036 at $r=200$, while CCA requires 147 latent features to explain 95% of the linearly explained variance; the paper reads this as evidence that the intrinsic dimension of this input-output map is below the linear estimate. The paper also claims that combining KL and covariance regularization yields more disentangled latent codes and better generative samples than either regularizer alone.

Load-bearing premise

The load-bearing premise is that fixing the latent code distribution to the same standard Gaussian for every input still lets the decoder produce outputs that faithfully depend on the input; if the decoder actually needs to see the input to capture that dependence, the claimed conditional model $p(y \mid x)$ does not follow from the training objective.

Editorial extensions

If this is right

  • At $r=50$ the VED reconstructs 323 pressure outputs from 1,475 transmissivity inputs with a normalized test MSE around 0.05, so a nonlinear surrogate with a 50-dimensional code can stand in for the high-dimensional map.
  • Increasing the latent dimension from 50 to 100 reduces reconstruction MSE by about 22 percent, while increasing from 100 to 200 reduces it by only about 7 percent, so for this dataset most of the usable information already fits in roughly 100 codes.
  • Small nonzero $\beta$ and $\lambda$ reduce the KL divergence and off-diagonal covariance of latent codes without raising MSE, indicating that disentanglement regularization does not force a reconstruction-accuracy tradeoff in the tested range.
  • Decoding random Gaussian noise through a model trained with $\beta$ and $\lambda$ produces output distributions closer to the test-data marginals, so the trained decoder can be used as a generator for synthetic pressure data.

Reading between the lines

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

  • Editorial inference: because the prior $p(z \mid x)$ is fixed to $\mathcal{N}(0,I_r)$ independently of $x$, the formal generative model $p_\theta(y \mid x)$ integrates out $x$ and is not $x$-dependent; the encoder is what injects input dependence, so the model is best understood as a regularized autoencoder whose decoder doubles as a marginal generator rather than as a strictly fitted conditional
  • Editorial inference: the $r=50$ result is demonstrated on one groundwater dataset; testing the same architecture on other elliptic PDE input-output maps, such as heat conductivity to temperature fields, would show whether fifty codes suffice generally or only for this benchmark.
  • Editorial inference: replacing the Map2Grid preprocessing with a graph-convolutional encoder acting directly on the irregular mesh would test whether the learned low-dimensional representation depends on the input parametrization rather than only on the underlying physics.
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

3 major / 5 minor

Summary. The paper proposes a Variational Encoder-Decoder (VED) framework for learning low-dimensional representations of high-dimensional input–output relations in physical systems, and applies it to a groundwater flow model where the input is a 1475-dimensional log-transmissivity field and the output is a 323-dimensional pressure response. The method combines a β-weighted KL divergence with a DIP-VAE-II covariance penalty, and experiments compare latent dimensions r = 50, 100, 150, 200 and regularization weights β, λ ∈ {0, 0.01, 0.1}. The central claims are that the VED models the conditional distribution p(y|x), that it achieves accurate reconstruction with r as low as 50, and that joint β–λ regularization improves latent disentanglement and generative quality when decoding Gaussian noise.

Significance. If the framework worked as described, it would be a useful contribution to nonlinear dimensionality reduction for physics-based models, combining a variational objective with an explicit disentanglement penalty and demonstrating the approach on a nontrivial groundwater flow dataset. The paper's empirical design has strengths: the use of CCA to motivate the range of latent dimensions, the residual-convolution encoder architecture, and the systematic sweep over β, λ, and training-set size are all sensible. However, the central derivation is invalid as written: with the prior p(z|x) = N(0, Ir), the generative distribution p_θ(y|x) is independent of x, so the derived ELBO is not a bound on the intended conditional log-likelihood. In addition, the headline empirical comparison is based on selecting the best test-set epoch and hyperparameters, which introduces optimistic bias and invalidates the claim that r = 50 yields no significant loss. These are load-bearing issues for the paper's main contribution, so the current version is not acceptable.

major comments (3)
  1. [Section 2.1, Eqs. (1)–(4)] The model definition makes the central derivation unsound. Eq. (1) defines p_θ(y, z|x) = p_θ(y|z) p(z|x), and the text after Eq. (4) states 'we have chosen for simplicity the prior p(z | x) ∼ N(0, Ir).' With this choice, p_θ(y|x) = ∫ p_θ(y|z) N(0, Ir) dz, which is independent of x. Therefore the ELBO in Eq. (2) is a bound on the marginal log-likelihood log p_θ(y), not on the conditional log-likelihood log p_θ(y|x). Consequently, the claimed equivalence to minimizing E_{p(x,y)} D_KL(q_φ(z|x) ∥ p_θ(z|x,y)) plus E_{p(x)} D_KL(p(y|x) ∥ p_θ(y|x)) is incorrect for the model as stated. The loss may still be a reasonable regularized conditional autoencoder objective, but it is not a variational bound on the conditional distribution the paper claims to model. To support the central claim, the prior would need to depend on x or the decoder would need to depend on x, either of which would change the method and the experiments.
  2. [Section 4.1 and Table 1] The reported reconstruction numbers are optimized on the test set, so the main empirical comparison is not valid. Section 4.1 states that 'we compute the MSE and KLD metrics for the test data after each epoch and store the set of parameters θ, φ that result in the best test MSE.' This selects the best of 100 epochs on the same 5,000-sample test set used for the final numeric reports, which introduces a strong optimistic bias. Hyperparameters β and λ were also chosen from preliminary experiments on the same data, and no repeated seeds or error bars are reported. Thus the headline claim that r = 50 gives performance 'without a significant loss of reconstruction accuracy' is unsupported; for example, the best entries in Table 1 are MSE 0.0484 at r = 50 (β = 0.1, λ = 0.1), 0.0383 at r = 100 (β = 0.01, λ = 0.1), and 0.0350 at r = 200 (β = 0.01, λ = 0.1), a gap of roughly 21–28% that could be a selection artifact. The experiments need to be redone with a validation set for epoch selection and multiple seeds to report mean and variance.
  3. [Section 4.3.1] The decoding-Gaussian-noise experiments do not validate the claimed conditional generative model. Figure 7 shows that the distribution of synthetic outputs from the decoder fed with N(0, Ir) noise resembles the marginal test-data distribution when β and λ are tuned. As the authors themselves note, the COV(φ) penalty pushes the aggregate encoding distribution q_φ(z) toward N(0, Ir); hence decoding noise from the prior will produce plausible marginal samples by construction. These results therefore provide evidence about the marginal output distribution p_θ(y), not about the conditional distribution p_θ(y|x), which is the paper's main claim. The interpretation should be corrected or the conditional claim should be tested by conditioning on specific x.
minor comments (5)
  1. [Equation (9)] Equation (9) writes β KLD(θ), but KLD is defined in Eq. (6) as a function of φ only (KLD(φ)); the argument should be φ, not θ.
  2. [Equation (5)] In Eq. (5), the noise ε is written as ε ∼ N(0, I_m), but ε is added to the r-dimensional latent code z = g_φ(x) + ε ⊙ exp(½ h_φ(x)); it should be N(0, I_r).
  3. [Section 4.2, text after Table 1] The text states that 'the smallest MSE values for each choice of r occur with λ,' which is contradicted by Table 1: for r = 50 and β = 0.1, λ = 0 gives MSE 0.0496 while λ = 0.1 gives 0.0504; for r = 200 and β = 0.1, λ = 0 gives 0.0427 while λ = 0.1 gives 0.0486. The claim should be qualified or corrected.
  4. [Section 3.2] The statement that approaches using 'aggressive striding and large feature expansion' work best is not supported by any ablation or quantitative comparison in the paper; either provide supporting data or soften the claim.
  5. [Section 3.1] The sentence 'The Hanford site dataset's log-transmissivity input consists of 1475 corresponding to the FV model's irregular quadrilateral grid' is incomplete and should be rephrased.

Circularity Check

1 steps flagged · score 6.0 of 10

Reported r=50 reconstruction accuracy is a test-set-minimized statistic, not an independent prediction.

  1. fitted input called prediction [Section 4.1 (Training Strategy) and Table 1]
    "For each experiment, we compute the MSE and KLD metrics for the test data after each epoch and store the set of parameters θ, φ that result in the best test MSE. Table 1 presents the best MSE and the corresponding KLD, and we highlight in bold the best MSE and corresponding KLD for each choice of r."

    The reported 'best test MSE' is by construction the minimum over 100 epochs of the MSE evaluated on the 5,000-sample test set. Because the test set is used to select the epoch (and hyperparameters were chosen after preliminary testing on the same data), the Table 1 numbers are optimistically biased minima, not unbiased estimates of a fixed model. The abstract's central claim—'as low as r = 50 latent dimensions without a significant loss of reconstruction accuracy'—is a comparison of these test-set minima (e.g., 0.0484 for r=50 vs 0.0383 for r=100 at β=λ=0.01). Thus the empirical prediction reduces to a fitted test-set statistic rather than an independent evaluation.

full rationale

The variational encoder-decoder derivation itself is standard: the ELBO, the reparameterized MSE, the closed-form KLD, and the covariance penalty follow from the stated Gaussian assumptions and are not self-referential. No load-bearing self-citation chain appears; the groundwater model citation [33] is an external data source, and the regularization citations [21, 25] are independent prior work. However, the paper's headline empirical claim—that r=50 latent dimensions reconstruct the test response 'without a significant loss'—is undermined by a fitted-input-as-prediction step: the test set is used both to select the best epoch (and, per Sections 2.2 and 4.1, to guide hyperparameter choices) and to report the final MSE. The reported numbers are therefore minima over test-set evaluations, so the r=50 vs r=100 vs r=200 comparison is a comparison of optimistically biased statistics, not an independent prediction. Additionally, the finding that the COV penalty makes the aggregate latent covariance closer to identity is tautological given Eq. (8), though this is secondary. The x-independent prior p(z|x)=N(0,I) also makes the generative conditional pθ(y|x) independent of x, which is a correctness risk rather than a circular reduction. Overall, the central empirical claim partially reduces to a test-set fit, warranting a score of 6.

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

No new physical entities are introduced; the latent code is a standard construct. The main hand-tuned quantities are the regularization weights beta and lambda and the latent dimension r. The p(z|x)=N(0,I) choice is an ad hoc simplification that undermines the conditional interpretation.

free parameters (3)
  • beta (KL regularization weight) = 0, 0.01, 0.1
    Tuned by grid search and test-based evaluation; controls the balance between reconstruction and KL regularization. The paper treats beta as sigma^2 but does not estimate sigma from data.
  • lambda (covariance penalty weight) = 0, 0.01, 0.1
    Weight for the DIP-VAE-II covariance penalty; the paper states lambda is not formally motivated and is chosen by test performance.
  • latent dimension r = 50, 100, 150, 200
    Range chosen from the CCA 95% explained-variance threshold at r=147; final headline result uses r=50. This is a hand-chosen model hyperparameter.
assumptions (3)
  • standard math The ELBO derivation follows from Jensen's inequality applied to log p(y|x).
    Section 2.1; the variational bound itself is standard, though the chosen prior makes the bound not conditional on x.
  • domain assumption The true relationship between log-transmissivity and pressure is well approximated by a Gaussian encoder and decoder with a standard Gaussian prior on latent codes.
    Section 2.1 and Section 3; the Gaussian forms for q_phi(z|x) and p_theta(y|z) are chosen for tractability, with no independent verification that the latent posterior is Gaussian.
  • ad hoc to paper The latent prior p(z|x) can be set to N(0,Ir), independent of x.
    Section 2.1, 'we have chosen for simplicity the prior p(z|x) ~ N(0,Ir)'. This choice makes p(y|x) independent of x, breaking the claimed conditional generative model; it is an ad hoc simplification rather than a justified modeling assumption.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Variational Encoder-Decoders for Learning Latent Representations of Physical Systems." pith.science (2026). https://pith.science/paper/STHCZA7N

@misc{pith2026241205175,
  author       = {Pith},
  title        = {Pith review of: Variational Encoder-Decoders for Learning Latent Representations of Physical Systems},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/STHCZA7N}},
  note         = {Machine review of arXiv:2412.05175}
}
abstract

We present a deep-learning Variational Encoder-Decoder (VED) framework for learning data-driven low-dimensional representations of the relationship between high-dimensional parameters of a physical system and the system's high-dimensional observable response. The framework consists of two deep learning-based probabilistic transformations: An encoder mapping parameters to latent codes and a decoder mapping latent codes to the observable response. The hyperparameters of these transformations are identified by maximizing a variational lower bound on the log-conditional distribution of the observable response given parameters. To promote the disentanglement of latent codes, we equip this variational loss with a penalty on the off-diagonal entries of the aggregate distribution covariance of codes. This regularization penalty encourages the pushforward of a standard Gaussian distribution of latent codes to approximate the marginal distribution of the observable response. Using the proposed framework we successfully model the hydraulic pressure response at observation wells of a groundwater flow model as a function of its discrete log-hydraulic transmissivity field. Compared to the canonical correlation analysis encoding, the VED model achieves a lower-dimensional latent representation, with as low as $r = 50$ latent dimensions without a significant loss of reconstruction accuracy. We explore the impact of regularization on model performance, finding that KL-divergence and covariance regularization improve feature disentanglement in latent space while maintaining reconstruction accuracy. Furthermore, we evaluate the generative capabilities of the regularized model by decoding random Gaussian noise, revealing that tuning both $\beta$ and $\lambda$ parameters enhances the quality of the generated observable response data.

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Pic2Spec: Generative Modeling Reconstructs Single Cell Raman Fingerprints from Brightfield Images

    physics.optics 2026-07 conditional novelty 6.0 of 10

    A dual-decoder variational autoencoder reconstructs single-cell Raman spectra from brightfield images with ~98% cosine similarity and discriminates bacterial transgenic states at 88% accuracy.

Reference graph

Works this paper leans on

35 extracted references · 24 canonical work pages · cited by 1 Pith paper

  1. [1]

    Geometric disentanglement for generative latent shape models

    Tristan Aumentado-Armstrong, Stavros Tsogkas, Allan Jepson, and Sven Dickinson. Geometric disentanglement for generative latent shape models. In 2019 IEEE/CVF International Conference on Computer Vision (ICCV) , pages 8180–8189, 2019. doi: 10.1109/ICCV.2019.00827

    work page arXiv 2019

  2. [2]

    Non-linear dimensionality reduction with a varia- tional encoder decoder to understand convective processes in climate models

    Gunnar Behrens, Tom Beucler, Pierre Gentine, Fernando Iglesias-Suarez, Michael Pritchard, and Veronika Eyring. Non-linear dimensionality reduction with a varia- tional encoder decoder to understand convective processes in climate models. Jour- nal of Advances in Modeling Earth Systems , 14(8):e2022MS003130, 2022. doi: 10.1029/2022MS003130. e2022MS003130 2...

    work page doi:10.1029/2022ms003130 2022

  3. [3]

    Representation learning: A review and new perspectives, 2014

    Yoshua Bengio, Aaron Courville, and Pascal Vincent. Representation learning: A review and new perspectives, 2014. URL https://arxiv.org/abs/1206.5538

    arXiv 2014

  4. [4]

    Isolating sources of disentanglement in variational autoencoders

    Ricky TQ Chen, Xuechen Li, Roger B Grosse, and David K Duvenaud. Isolating sources of disentanglement in variational autoencoders. Advances in neural informa- tion processing systems, 31, 2018

    work page 2018

  5. [5]

    Transient inverse calibration of Hanford site-wide ground- water model to Hanford operational impacts – 1943 to 1996

    Charles R Cole, Marcel P Bergeron, Signe K Wurstner, Paul D Thorne, Samuel Orr, and Mathew I Mckinley. Transient inverse calibration of Hanford site-wide ground- water model to Hanford operational impacts – 1943 to 1996. Technical report, Pacific Northwest National Laboratory (PNNL), Richland, Washington, United States, 2001

    work page 1943

  6. [6]

    Tutorial on variational autoencoders, 2021

    Carl Doersch. Tutorial on variational autoencoders, 2021. URL https://arxiv.org/ abs/1606.05908

    arXiv 2021

  7. [7]

    Structured dis- entangled representations

    Babak Esmaeili, Hao Wu, Sarthak Jain, Alican Bozkurt, Narayanaswamy Siddharth, Brooks Paige, Dana H Brooks, Jennifer Dy, and Jan-Willem Meent. Structured dis- entangled representations. In The 22nd International Conference on Artificial Intel- ligence and Statistics , pages 2525–2534. PMLR, 2019

    work page 2019

  8. [8]

    Harry R. Glahn. Canonical correlation and its relationship to discriminant analysis and multiple regression. Journal of Atmospheric Sciences , 25(1):23 – 31, 1968. doi: 10.1175/1520-0469(1968)025⟨0023:CCAIRT⟩2.0.CO;2

    doi:10.1175/1520-0469(1968)025 1968

Show all 35 references
  1. [9]

    Deep residual learning for image recognition

    Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. Deep residual learning for image recognition. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), June 2016

    2016

  2. [10]

    beta-V AE: Learning basic vi- sual concepts with a constrained variational framework

    Irina Higgins, Loic Matthey, Arka Pal, Christopher Burgess, Xavier Glorot, Matthew Botvinick, Shakir Mohamed, and Alexander Lerchner. beta-V AE: Learning basic vi- sual concepts with a constrained variational framework. In International Conference on Learning Representations, 2017. 19

    2017

  3. [11]

    G. E. Hinton and R. R. Salakhutdinov. Reducing the dimensionality of data with neural networks. Science, 313(5786):504–507, 2006. doi: 10.1126/science.1127647

    2006 doi

  4. [12]

    Batch normalization: accelerating deep network training by reducing internal covariate shift

    Sergey Ioffe and Christian Szegedy. Batch normalization: accelerating deep network training by reducing internal covariate shift. In Proceedings of the 32nd International Conference on International Conference on Machine Learning - Volume 37 , ICML’15, page 448–456. JMLR.org, 2015

    2015

  5. [13]

    Disentangling generative factors of phys- ical fields using variational autoencoders

    Christian Jacobsen and Karthik Duraisamy. Disentangling generative factors of phys- ical fields using variational autoencoders. Frontiers in Physics, 10:890910, 2022

    2022

  6. [14]

    Deep encoder, shallow decoder: Reevaluating non-autoregressive machine translation

    Jungo Kasai, Nikolaos Pappas, Hao Peng, James Cross, and Noah Smith. Deep encoder, shallow decoder: Reevaluating non-autoregressive machine translation. In International Conference on Learning Representations , 2021. URL https:// openreview.net/forum?id=KpfasTaLUpq

    2021

  7. [15]

    Disentangling by factorising

    Hyunjik Kim and Andriy Mnih. Disentangling by factorising. In International con- ference on machine learning , pages 2649–2658. PMLR, 2018

    2018

  8. [16]

    γ-V AE: Curvature reg- ularized variational autoencoders for uncovering emergent low dimensional geometric structure in high dimensional data

    Jason Z Kim, Nicolas Perrin-Gilbert, Erkan Narmanli, Paul Klein, Christopher R Myers, Itai Cohen, Joshua J Waterfall, and James P Sethna. γ-V AE: Curvature reg- ularized variational autoencoders for uncovering emergent low dimensional geometric structure in high dimensional da...

    2024 doi arXiv

  9. [17]

    Kingma and Jimmy Ba

    Diederik P. Kingma and Jimmy Ba. Adam: A method for stochastic optimization, 2017

    2017

  10. [18]

    Kingma and Max Welling

    Diederik P. Kingma and Max Welling. Auto-encoding variational bayes.arXiv preprint arXiv:1312.6114, 2022. doi: 10.48550/arXiv.1312.6114

    2022 doi arXiv

  11. [19]

    Kipf and Max Welling

    Thomas N. Kipf and Max Welling. Semi-supervised classification with graph convo- lutional networks, 2017

    2017

  12. [20]

    Alex Krizhevsky, Ilya Sutskever, and Geoffrey E. Hinton. Imagenet classification with deep convolutional neural networks. In Proceedings of the 25th International Conference on Neural Information Processing Systems - Volume 1 , NIPS’12, page 1097–1105, Red Hook, NY, USA, 2012....

    2012

  13. [21]

    Variational infer- ence of disentangled latent concepts from unlabeled observations

    Abhishek Kumar, Prasanna Sattigeri, and Avinash Balakrishnan. Variational infer- ence of disentangled latent concepts from unlabeled observations. In International Conference on Learning Representations , 2018. URL https://openreview.net/ forum?id=H1kG7GZAW

    2018

  14. [22]

    Lecun, L

    Y. Lecun, L. Bottou, Y. Bengio, and P. Haffner. Gradient-based learning applied to document recognition. Proceedings of the IEEE , 86(11):2278–2324, 1998. doi: 10.1109/5.726791. 20

    1998 doi

  15. [23]

    Information con- straints on auto-encoding variational bayes.Advances in neural information processing systems, 31, 2018

    Romain Lopez, Jeffrey Regier, Michael I Jordan, and Nir Yosef. Information con- straints on auto-encoding variational bayes.Advances in neural information processing systems, 31, 2018

    2018

  16. [24]

    Stacked convolu- tional auto-encoders for hierarchical feature extraction

    Jonathan Masci, Ueli Meier, Dan Cire¸ san, and J¨ urgen Schmidhuber. Stacked convolu- tional auto-encoders for hierarchical feature extraction. In Timo Honkela, W lodzis law Duch, Mark Girolami, and Samuel Kaski, editors, Artificial Neural Networks and Ma- chine Learning – ICA...

    2011

  17. [25]

    Disentangling disentanglement in variational autoencoders

    Emile Mathieu, Tom Rainforth, N Siddharth, and Yee Whye Teh. Disentangling disentanglement in variational autoencoders. In Kamalika Chaudhuri and Ruslan Salakhutdinov, editors, Proceedings of the 36th International Conference on Machine Learning, volume 97 of Proceedings of Ma...

    2019

  18. [26]

    Linear and nonlinear dimensionality reduction from fluid me- chanics to machine learning

    Miguel A Mendez. Linear and nonlinear dimensionality reduction from fluid me- chanics to machine learning. Measurement Science and Technology , 34(4):042001, jan 2023. doi: 10.1088/1361-6501/acaffe. URL https://dx.doi.org/10.1088/ 1361-6501/acaffe

    2023 doi

  19. [27]

    On the difficulty of train- ing recurrent neural networks

    Razvan Pascanu, Tomas Mikolov, and Yoshua Bengio. On the difficulty of train- ing recurrent neural networks. In Proceedings of the 30th International Confer- ence on International Conference on Machine Learning - Volume 28 , ICML’13, page III–1310–III–1318. JMLR.org, 2013

    2013

  20. [28]

    Simple and effective vae training with calibrated decoders

    Oleh Rybkin, Kostas Daniilidis, and Sergey Levine. Simple and effective vae training with calibrated decoders. In Marina Meila and Tong Zhang, editors, Proceedings of the 38th International Conference on Machine Learning , volume 139 of Proceedings of Machine Learning Research...

    2021

  21. [29]

    Assessing the interplay of shape and physical parameters by nonlinear dimensionality reduction methods

    Andrea Serani, Danny D’Agostino, Emilio Campana, and Matteo Diez. Assessing the interplay of shape and physical parameters by nonlinear dimensionality reduction methods. In 32nd Symposium on Naval Hydrodynamics Hamburg, Germany , 08 2018

    2018

  22. [30]

    Learning structured out- put representation using deep conditional generative models

    Kihyuk Sohn, Honglak Lee, and Xinchen Yan. Learning structured out- put representation using deep conditional generative models. In C. Cortes, N. Lawrence, D. Lee, M. Sugiyama, and R. Garnett, editors, Advances in Neural Information Processing Systems , volume 28. Curran Assoc...

  23. [31]

    Unsupervised geomet- ric disentanglement via CF AN-V AE

    Norman Joseph Tatro, Stefan C Schonsheck, and Rongjie Lai. Unsupervised geomet- ric disentanglement via CF AN-V AE. In ICLR 2021 Workshop on Geometrical and 21 Topological Representation Learning, 2021. URL https://openreview.net/forum? id=shaSYHQSEVf

    2021

  24. [32]

    Recent advances in autoencoder-based representation learning

    Michael Tschannen, Olivier Bachem, and Mario Lucic. Recent advances in autoencoder-based representation learning. arXiv preprint arXiv:1812.05069 , 2018

    2018 arXiv

  25. [33]

    Barajas-Solano, and Alexandre M

    Yu-Hong Yeung, David A. Barajas-Solano, and Alexandre M. Tartakovsky. Gaussian process regression and conditional karhunen-lo` eve models for data assimilation in in- verse problems. Journal of Computational Physics , 502:112788, 2024. ISSN 0021-9991. doi: 10.1016/j.jcp.2024.112788

    2024

  26. [34]

    Zeiler and Rob Fergus

    Matthew D. Zeiler and Rob Fergus. Visualizing and understanding convolutional networks. In David Fleet, Tomas Pajdla, Bernt Schiele, and Tinne Tuytelaars, editors, Computer Vision – ECCV 2014 , pages 818–833, Cham, 2014. Springer International Publishing. ISBN 978-3-319-10590-...

    2014

  27. [2015]

    URL https://proceedings.neurips.cc/paper_files/paper/2015/file/ 8d55a249e6baa5c06772297520da2051-Paper.pdf

    2015

Pith tools

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