Pith. sign in

REVIEW 3 major objections 4 minor 48 references

Neighbour-Driven Gaussian Process Variational Autoencoders for Scalable Structured Latent Modelling

T0 review · 3 major / 4 minor · reviewed 2026-08-07 · deepseek-v4-flash

Pith's one-line read This paper claims that replacing the dense Gaussian-process prior over latent variables with a nearest-neighbour-truncated prior makes Gaussian-process variational autoencoders scalable and kernel-flexible, competitive with or better than…

desk verdict A solid, reproducible application of Vecchia-style nearest-neighbour conditioning to GPVAE latent spaces, with an overstated flexible-kernel claim and a hand-wavy HPA derivation. read the letter →

arxiv 2505.16481 v3 pith:N6A7PEFI submitted 2025-05-22 cs.LG stat.ML

classification cs.LGstat.ML MSC 68T0762M30
keywords Gaussianprocessvariationalautoencodernearestneighbourscalableinferencestructuredlatentmodelssparseprecisionapproximationhierarchicalpriormini-batchtrainingdataimputation
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper sets out to remove the main scalability bottleneck of Gaussian Process Variational Autoencoders (GPVAEs): the dense $N\times N$ covariance matrix of the GP prior, which makes exact inference $O(N^3)$ and blocks mini-batching. Its central claim is that a local-neighbourhood truncation of that prior — keeping only the $H$ nearest neighbours of each auxiliary location — preserves the latent correlations that matter for structured data while cutting per-step cost to $O(N_b H^3)$. Two concrete approximations are proposed and tested: Hierarchical Prior Approximation (HPA), which switches off interactions between non-neighbours through a latent selection variable, and Sparse Precision Approximation (SPA), which factorises the prior into conditional distributions on the $H$ nearest neighbours. On representation learning, imputation, and conditional generation tasks (including a spatial dataset with about 140,000 points), the paper reports that both variants match or improve on full-batch and inducing-point GPVAE baselines in predictive error and training speed.

What carries the argument

The load-bearing object is the $H$-neighbour truncation of the latent GP prior, in two complementary forms. HPA (Hierarchical Prior Approximation) imposes sparsity on the covariance: a binary selection vector $w$ activates only neighbourhood-sized blocks $D_w K_{XX} D_w$, so each mini-batch's KL uses an $H\times H$ covariance. SPA (Sparse Precision Approximation) imposes sparsity on the precision: the joint GP prior is factorised via the probability chain rule and each factor $p(z_j \mid z_{n(j)})$ conditions only on the $H$ nearest neighbours of $x_j$ among previously ordered points, giving a sparse precision/Cholesky structure. Both replace the dense $N\times N$ covariance or precision with $H\times H$ pieces, so the per-iteration cost is $O(L N_b H^3)$ once nearest neighbours are precomputed, and both reduce to the full GPVAE ELBO when $H=N$ and to a plain VAE when $H=0$. The same local conditioning is used at prediction time, so unseen locations are handled by their $H$ neighbours only.

What would settle it

Train GPVAE-SPA and GPVAE-HPA on a synthetic temporal dataset whose latent trajectories are drawn from a GP with an RBF kernel whose lengthscale is several times the observation window, so every point correlates strongly with every other point. If, with $H$ fixed at a small value such as 10, the latent-trajectory reconstruction RMSE is far worse than a full-batch GPVAE, while the same models match the full-batch on short-lengthscale trajectories, the local-adjacency premise is false for long-range dependence. A quantitative precursor would be the KL divergence between the true $N$-dimensional GP prior and the $H$-neighbour approximation, which should stay large for small $H$ under a long lengthscale.

Watch

Extended reading notes

Core claim

The paper's contention is that the GP prior over latent variables in a GPVAE can be approximated by a neighbour-driven sparsity pattern without losing the structured dependence the prior is there to encode. HPA multiplies the dense GP covariance by a binary diagonal selection matrix $D_w$, so that only latent variables belonging to the same selected neighbourhood interact, yielding an ELBO whose KL term splits into $H\times H$ sub-blocks. SPA instead writes the prior as a product of conditional densities $p(z_j \mid z_{n(j)})$, where $n(j)$ collects the $H$ nearest neighbours of $x_j$ among earlier points, which is a sparse Cholesky-type approximation of the precision matrix; its ELBO again separates into mini-batches. In both cases the training objective is amortised over mini-batches and accepts arbitrary kernels, and prediction at an unseen location $x_*$ conditions only on the $H$ nearest neighbours of $x_*$. The paper presents experiments on moving-ball trajectories, rotated handwritten-digit sequences, physics-simulation action series, and two geostatistical datasets, reporting that the two approximations approach or exceed the accuracy of full-batch GPVAE baselines and of inducing-point models that use comparable or larger budgets, while training faster.

Load-bearing premise

The load-bearing premise is that the $H$ nearest neighbours of each point in the auxiliary space carry essentially all of the GP-prior dependence that matters, so that correlations to everything further away can be zeroed out without distorting the latent structure.

Editorial extensions

If this is right

  • Training scales to large structured datasets: with $H$ fixed at, e.g., 20, the per-mini-batch cost $O(N_b H^3)$ is independent of total $N$, and the paper demonstrates a 140,000-point spatial imputation running in seconds per epoch.
  • Arbitrary kernels become usable in the latent prior, including RBF, Matérn, and Cauchy kernels, because the approximation needs only pairwise kernel evaluations on neighbour sets and avoids the low-rank or state-space kernel restrictions of earlier GPVAE variants; the paper's experiments lean on this flexibility.
  • Inducing points are no longer the resolution mechanism: on the moving-ball data, the paper reports that the sparse-precision variant reaches good reconstruction with about one sixth of the trajectory points as neighbours, where an inducing-point baseline needed roughly half the trajectory as inducing inputs to approach the full-batch result.
  • With $H=N$ the objectives recover the full-batch GPVAE ELBO, and with $H=0$ the SPA objective degenerates to a standard VAE, so the method forms a continuous family from factorised to fully structured latent priors.
  • Conditional generation at an unseen timestamp or location is cheap: prediction at $x_*$ uses only the $H$ nearest neighbours of $x_*$, a small encoder pass, and one $H\times H$ linear solve.

Reading between the lines

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

  • The choice of distance metric for 'nearest' is a hidden design parameter; the paper uses Euclidean distance throughout. Swapping in a kernel-induced or manifold-aware distance could change both HPA and SPA behaviour on data whose local structure is curved, such as rotation manifolds, and would be a direct test of how much the neighbourhood definition matters.
  • Because both approximations interpolate between a factorised VAE at $H=0$ and the exact GP prior at $H=N$, the gap between their curves as $H$ grows is a measurable diagnostic of how locally structured a dataset actually is; datasets where small-$H$ performance is already near the full-batch result are precisely those where the method's premise holds.
  • The same trick should transfer to other structured priors: any latent model whose prior is a Gaussian Markov random field or a graph GP could use graph-adjacency neighbourhoods instead of coordinate distance, extending the scalability claim beyond spatial and temporal auxiliaries.
  • The consistent small edge of SPA over HPA in the paper's tables suggests that sparsifying the precision (conditional independence) may be a more faithful local approximation than sparsifying the covariance (zeroing non-neighbour covariances); comparing the two on a long-range-kernel dataset would show whether that ordering generalises.
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 / 4 minor

Summary. The paper proposes two neighbour-driven approximations to the Gaussian-process prior in GPVAEs: Hierarchical Prior Approximation (HPA), Eq. (10), and Sparse Precision Approximation (SPA), Eq. (13). Both restrict prior dependencies to the H nearest neighbours of each auxiliary location, giving O(N_b H^3) training complexity and enabling mini-batch inference with arbitrary kernel functions. The authors evaluate the methods on latent trajectory reconstruction, rotated-MNIST imputation and generation, MuJoCo conditional generation, and two geostatistical datasets, reporting improvements over several GPVAE baselines in prediction quality and wall-clock time.

Significance. If the claims hold, the paper provides a practical alternative to inducing-point-based GPVAEs for structured latent modelling on data with local dependence, with an open-source implementation and a careful set of benchmarks. The SPA construction is a standard Vecchia-style approximation and is coherent as a scalable surrogate for the full GP prior. The reported experiments are extensive, use 10 random trials, and include both small and large datasets, which makes the empirical contribution valuable. The main caveats are that the HPA objective is not derived as a valid ELBO for the stated hierarchical model, and that the nearest-neighbour premise restricts the class of kernels for which the approximation is statistically faithful.

major comments (3)
  1. [Appendix B.1, Eq. (10)] The displayed objective L_HPA is not presented as a valid ELBO for the hierarchical model in Eqs. (6)-(9). In the derivation, the full likelihood term E_{p(w)} E_{q(Z|w)} log p(Y|Z) is replaced by N/|I| times a sum over the mini-batch I, and the full KL term is replaced by an average over neighbourhood KL divergences. When the neighbourhoods n(i) overlap, the KL term counts overlapping latent variables multiple times, while the likelihood term ignores selected neighbours that are not in I. In addition, q(z_i|w) from Eq. (8) is replaced by q(z_i|y_i), which is not justified. Thus L_HPA is, for finite H, a block-wise surrogate objective rather than a lower bound on log p(Y). The statement that H=N recovers Eq. (4) only establishes agreement in the limit and does not validate the finite-H objective. The authors should either define a variational family and a proper distribution p(w) for which Eq. (10) is a genuine ELBO, or explicitly present HPA as an approximate training objective and analyse the bias introduced by the block approximation.
  2. [Abstract, Section 3.2, Section 3.3, Table 15] The claims that the method 'preserves essential latent dependencies' and supports 'arbitrary kernels' are scoped too broadly. The H-neighbour truncation in Eqs. (10) and (13) is a conditional-independence assumption: it is faithful only when kernel-induced dependence is dominated by short-range correlations. For a periodic kernel such as k(x,x') = sigma^2 exp(-2 sin^2(pi(x-x')/T)/ell^2) with period T larger than the H-neighbour window, the nearest neighbours are nearly uncorrelated while points one period apart are strongly correlated; the H-neighbour prior and the predictive rule in Section 3.3 would discard the dominant dependencies. Every experiment in the paper uses RBF, Matérn-3/2, or Cauchy kernels on temporal or spatial grids with monotone distance decay, so this failure regime is not exercised. The claims should be restricted to kernels with dominantly local correlation structure, or a long-range/periodic-kernel experiment should be added to demonstrate where the approximation breaks down.
  3. [Introduction and Table 1] The central empirical claim that the approach 'improves both predictive accuracy and training speed compared to existing GPVAE baselines' is contradicted by the Rotated-MNIST corrupted-frame results in Table 1. There, MGPV AE achieves NLL 0.090 +/- 0.001 versus 0.095 +/- 0.000 for HPA and 0.096 +/- 0.000 for SPA, RMSE 0.197 +/- 0.001 versus 0.199 and 0.202, and a faster training time of 110.8 s/epoch versus 137.9 and 121.2 s/epoch. The paper acknowledges this in Section 5.2, but the Abstract and Section 1 still make an unqualified superiority claim. The superiority claims should be qualified to the settings and metrics where they actually hold.
minor comments (4)
  1. [Section 3.4] The complexity statement 'Locating the H nearest neighbours of each point takes O(HN) in the worst case' is dimensionally incomplete: a per-query cost of O(HN) gives O(H N^2) for all N points, and the Faiss-based approximate search does not have this worst-case guarantee as stated.
  2. [Section 3.2, SPA discussion] The sentence 'setting H = 0 will cause the model to degenerate into conventional VAEs' is only true when the marginal variances k(x_j, x_j) are 1; in general H=0 gives a VAE with a diagonal Gaussian prior with per-point variances. Please clarify.
  3. [Section 3.3] The predictive posterior is derived using the SPA-style conditional p(z_* | Z_n(*)); for HPA, whose approximate posterior is not defined through a chain factorisation, the same formula is used without justification. Please state explicitly that this is a shared approximation for both variants.
  4. [Appendix B.1 and Eq. (10)] The notation for the variational distribution is inconsistent: Eq. (10) writes E_{q(z_i|y_i)}, while the derivation uses E_{q(z_i|w)}. Please align these notations and define the marginal q(z_i) used in the final objective.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the neighbour-driven GPVAE derivations are explicit approximations, and the reported predictions are not fitted to the evaluated metrics.

full rationale

The paper's derivation chain starts from the standard GPVAE ELBO (Eq. 4), then introduces HPA through an explicit hierarchical selection variable and a mini-batch neighbour-based ELBO estimate (Eq. 10), and SPA through a Vecchia-style chained conditional factorisation (Eqs. 11-13). Each step is a defined approximation with stated assumptions; the predictive posterior in Section 3.3 follows from the GP conditional distribution given the H nearest neighbours, not from the test metrics. No parameter is fitted to the reported NLL or RMSE and then renamed as a prediction. The paper contains no self-citations; its foundational references (Vecchia, Datta, Wu et al., etc.) are external, and the NNGP locality premise is openly adopted as an assumption rather than smuggled in. The statement that the full-batch ELBO is recovered when H=N is a consistency check, not a circular reduction. The limitations note about Euclidean distance raises a scope/correctness concern for non-local kernels, but it is not a circular step. The derivation is therefore self-contained with respect to the claims it makes, and the empirical comparisons are genuine external evaluations.

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

The central contribution is an approximation strategy, not a new physical entity. The cost side is carried by two hand-chosen hyperparameters (H and beta) and by the locality and factorisation assumptions above. The approximate nature of HPA and SPA is the price paid for scalability.

free parameters (2)
  • H (neighbourhood size) = 5, 10, 20 depending on task
    Number of nearest neighbours used in HPA and SPA; chosen per dataset and directly controls cost and fidelity.
  • beta (ELBO trade-off) = 1.0, 1.5, 1500, 0.2 depending on task and model
    Weights the likelihood versus KL term; tuned separately for HPA and SPA in several experiments, with no sensitivity analysis reported.
assumptions (4)
  • domain assumption A GP prior on latent variables Z with kernel over auxiliary inputs X models structured correlations.
    Adopted from the GPVAE construction in Section 2.1, Eq. (1).
  • domain assumption Local adjacency in X captures essential latent dependence, so H nearest neighbours are enough.
    Section 3.2 states the key insight in terms of local adjacencies and the first law of geography; this motivates both HPA and SPA.
  • domain assumption A fully factorised encoder q_phi(Z|Y) is sufficient for amortized inference.
    Section 3.1: 'We choose to employ the standard encoder from (3)' to keep the model simple.
  • domain assumption SPA's chain with a fixed ordering and previous-neighbour conditionals is a valid GP approximation.
    Section 3.2 Eq. (12) relies on Vecchia (1988); the ordering itself is unspecified for spatial data.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Neighbour-Driven Gaussian Process Variational Autoencoders for Scalable Structured Latent Modelling." pith.science (2026). https://pith.science/paper/N6A7PEFI

@misc{pith2026250516481,
  author       = {Pith},
  title        = {Pith review of: Neighbour-Driven Gaussian Process Variational Autoencoders for Scalable Structured Latent Modelling},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/N6A7PEFI}},
  note         = {Machine review of arXiv:2505.16481}
}
read the original abstract

Gaussian Process (GP) Variational Autoencoders (VAEs) extend standard VAEs by replacing the fully factorised Gaussian prior with a GP prior, thereby capturing richer correlations among latent variables. However, performing exact GP inference in large-scale GPVAEs is computationally prohibitive, often forcing existing approaches to rely on restrictive kernel assumptions or large sets of inducing points. In this work, we propose a neighbour-driven approximation strategy that exploits local adjacencies in the latent space to achieve scalable GPVAE inference. By confining computations to the nearest neighbours of each data point, our method preserves essential latent dependencies, allowing more flexible kernel choices and mitigating the need for numerous inducing points. Through extensive experiments on tasks including representation learning, data imputation, and conditional generation, we demonstrate that our approach outperforms other GPVAE variants in both predictive performance and computational efficiency.

Figures

Figures reproduced from arXiv: 2505.16481 by the authors.

Figure 1
Figure 1. GPVAE places GP prior over the latent variables Z to model correlations in the structured data (xi , yi). Our approach is to approximate the dense GP prior by leveraging the associated neighbourhood xn(i) , yn(i)  . sequential, spatial, or other structured data. In these cases, it is crucial to model correlations among latent variables. A natural way to introduce such correlations is to adopt a Gaussian Process (GP… view at source ↗
Figure 2
Figure 2. Latent representation learning for the moving ball dataset. (a) The leftmost column shows frames overlaid and shaded by time. The orange reconstructed paths are obtained using M/H = 10. (b) The results of the standard VAE and two full-batch baselines are shown with the shaded bands. The error bars and the shades indicate ±1 standard deviation. p(y∗ | x∗, Y) at a new location x∗ only needs to consider its H nearest n… view at source ↗
Figure 3
Figure 3. (a) Corrupted frame imputation with around 60% missing pixels. (b) Missing frames generation on Rotated MNIST with around 60% missing frames. The red boxes indicate the missing frames in that sequence. We benchmark our models against diverse GPVAEs, mainly focusing on scalable models such as SVGPVAE (Jazbec et al., 2021), LVAE (Ramchandran et al., 2021), MGPVAE (Zhu et al., 2023), and SGPBAE (Tran et al., 2023). Som… view at source ↗
Figures from the paper (7 more)
Figure 5
Figure 5. Figure 5: Porosity imputation (i.e., the 1st channel) at the intermediate layer of SPE10 dataset using GPVAE models. the nearest neighbour principle should be particularly ef￾fective. The first dataset, Jura (Goovaerts, 1997), contains hundreds of data points, while the second d…
Figure 6
Figure 6. Figure 6: with 5 or 15 inducing points/nearest neighbours [PITH_FULL_IMAGE:figures/full_fig_p015_6.png]
Figure 7
Figure 7. Figure 7: illustrates additional imputation results for other digits in the dataset. Input VAE HI-VAE GPVAE-Diag (full batch) GPVAE-Band (full batch) MGPVAE GPVAE-HPA (ours) GPVAE-SPA (ours) Input VAE HI-VAE GPVAE-Diag (full batch) GPVAE-Band (full batch) MGPVAE GPVAE-HPA (ours)…
Figure 8
Figure 8. Figure 8: Missing frames generated from two unseen MNIST sequences. The red boxes indicate the missing frames. Additional results We provide additional NLL and RMSE results for our two proposed models in [PITH_FULL_IMAGE:figures/full_fig_p018_8.png]
Figure 9
Figure 9. Figure 9: Predictive mean for Cadmium on the Jura dataset test split (100 locations). Input RMSE=2.040 MOGP RMSE=0.983 VNNGP-H20 RMSE=1.506 VAE RMSE=0.665 HIVAE True Layer RMSE=1.967 SVGPVAE-M100 RMSE=0.882 SGPBAE-M20 RMSE=0.610 GPVAE-HPA-H20(ours) RMSE=0.601 GPVAE-SPA-H20(ours)…
Figure 10
Figure 10. Figure 10: Additional imputation results of SPE10. (top) The permeability along the y direction at the 26th layer; (middle) The permeability along the x direction at the 41st layer; (bottom) The porosity at the 22nd layer. 20 [PITH_FULL_IMAGE:figures/full_fig_p020_10.png]
Figure 11
Figure 11. Figure 11: 95% posterior credible intervals for a test MuJoCo sequence for all other dimensions with N = 1000. The red dots mark unobserved data points. 21 [PITH_FULL_IMAGE:figures/full_fig_p021_11.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

48 extracted references · 41 canonical work pages

  1. [1]

    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 global.max substring 't := if while FUNCTION format.date year duplicate empty "emp...

  2. [2]

    Doubly sparse variational Gaussian processes

    Adam, V., Eleftheriadis, S., Artemev, A., Durrande, N., and Hensman, J. Doubly sparse variational Gaussian processes. In International Conference on Artificial Intelligence and Statistics, pp.\ 2874--2884. PMLR, 2020

  3. [3]

    O., et al

    Allison, R., Stephenson, A., Pyzer-Knapp, E. O., et al. Leveraging locality and robustness to achieve massively scalable Gaussian process regression. Advances in Neural Information Processing Systems, 36, 2024

  4. [4]

    Ashman, M., So, J., Tebbutt, W., Fortuin, V., Pearce, M., and Turner, R. E. Sparse Gaussian process variational autoencoders. arXiv preprint arXiv:2010.10177, 2020

  5. [5]

    T., and Katzfuss, M

    Cao, J., Kang, M., Jimenez, F., Sang, H., Schaefer, F. T., and Katzfuss, M. Variational sparse inverse Cholesky approximation for latent Gaussian processes via double kullback-leibler minimization. In International Conference on Machine Learning, pp.\ 3559--3576. PMLR, 2023

  6. [6]

    P., Dalca, A., Saglietti, L., Listgarten, J., and Fusi, N

    Casale, F. P., Dalca, A., Saglietti, L., Listgarten, J., and Fusi, N. Gaussian process prior variational autoencoders. Advances in Neural Information Processing Systems, 31, 2018

  7. [7]

    E., Wilkinson, W

    Chang, P. E., Wilkinson, W. J., Khan, M. E., and Solin, A. Fast variational learning in state-space Gaussian process models. In 2020 IEEE 30th International Workshop on Machine Learning for Signal Processing (MLSP), pp.\ 1--6. IEEE, 2020

  8. [8]

    Stochastic gradient Hamiltonian Monte Carlo

    Chen, T., Fox, E., and Guestrin, C. Stochastic gradient Hamiltonian Monte Carlo . In International Conference on Machine Learning, pp.\ 1683--1691. PMLR, 2014

Show all 48 references
  1. [9]

    Christie, M. A. and Blunt, M. J. Tenth spe comparative solution project: A comparison of upscaling techniques. SPE Reservoir Evaluation & Engineering, 4 0 (04): 0 308--317, 2001

  2. [10]

    Nearest-neighbor sparse Cholesky matrices in spatial statistics

    Datta, A. Nearest-neighbor sparse Cholesky matrices in spatial statistics. Wiley Interdisciplinary Reviews: Computational Statistics, 14 0 (5): 0 e1574, 2022

  3. [11]

    O., and Gelfand, A

    Datta, A., Banerjee, S., Finley, A. O., and Gelfand, A. E. Hierarchical nearest-neighbor Gaussian process models for large geostatistical datasets. Journal of the American Statistical Association, 111 0 (514): 0 800--812, 2016 a

  4. [12]

    O., and Gelfand, A

    Datta, A., Banerjee, S., Finley, A. O., and Gelfand, A. E. On nearest-neighbor Gaussian process models for massive spatial data. Wiley Interdisciplinary Reviews: Computational Statistics, 8 0 (5): 0 162--171, 2016 b

  5. [13]

    Gp-vae: Deep probabilistic time series imputation

    Fortuin, V., Baranchuk, D., R \"a tsch, G., and Mandt, S. Gp-vae: Deep probabilistic time series imputation. In International Conference on Artificial Intelligence and Statistics, pp.\ 1651--1661. PMLR, 2020

  6. [14]

    R., Pleiss, G., Bindel, D., Weinberger, K

    Gardner, J. R., Pleiss, G., Bindel, D., Weinberger, K. Q., and Wilson, A. G. Gpytorch: Blackbox matrix-matrix Gaussian process inference with GPU acceleration. In Advances in Neural Information Processing Systems, volume 31, pp.\ 7587--7597, 2018

  7. [15]

    Geostatistics for natural resources evaluation, volume 483

    Goovaerts, P. Geostatistics for natural resources evaluation, volume 483. Oxford University Press, 1997

  8. [16]

    Spatio-temporal variational Gaussian processes

    Hamelijnck, O., Wilkinson, W., Loppi, N., Solin, A., and Damoulas, T. Spatio-temporal variational Gaussian processes. Advances in Neural Information Processing Systems, 34: 0 23621--23633, 2021

  9. [17]

    Hensman, J., Fusi, N., and Lawrence, N. D. Gaussian processes for big data. In Uncertainty in Artificial Intelligence, pp.\ 282. Citeseer, 2013

  10. [18]

    G., Filippone, M., and Ghahramani, Z

    Hensman, J., Matthews, A. G., Filippone, M., and Ghahramani, Z. MCMC for variationally sparse Gaussian processes. Advances in Neural Information Processing Systems, 28, 2015

  11. [19]

    P., Glorot, X., Botvinick, M

    Higgins, I., Matthey, L., Pal, A., Burgess, C. P., Glorot, X., Botvinick, M. M., Mohamed, S., and Lerchner, A. beta-vae: Learning basic visual concepts with a constrained variational framework. In International Conference on Learning Representations (ICLR), 2017. URL https://o...

  12. [20]

    Scalable Gaussian process variational autoencoders

    Jazbec, M., Ashman, M., Fortuin, V., Pearce, M., Mandt, S., and R \"a tsch, G. Scalable Gaussian process variational autoencoders. In International Conference on Artificial Intelligence and Statistics, pp.\ 3511--3519. PMLR, 2021

  13. [21]

    Billion-scale similarity search with GPUs

    Johnson, J., Douze, M., and J \'e gou, H. Billion-scale similarity search with GPUs . IEEE Transactions on Big Data, 7 0 (3): 0 535--547, 2019

  14. [22]

    and Katzfuss, M

    Kang, M. and Katzfuss, M. Correlation-based sparse inverse Cholesky factorization for fast Gaussian -process inference. Statistics and Computing, 33 0 (3): 0 56, 2023

  15. [23]

    and Guinness, J

    Katzfuss, M. and Guinness, J. A general framework for V ecchia approximations of Gaussian processes. Statistical Science, 36 0 (1), 2021

  16. [24]

    and Mnih, A

    Kim, H. and Mnih, A. Disentangling by factorising. In Dy, J. and Krause, A. (eds.), Proceedings of the 35th International Conference on Machine Learning, volume 80 of Proceedings of Machine Learning Research, pp.\ 2649--2658. PMLR, 10--15 Jul 2018

  17. [25]

    Kingma, D. P. and Welling, M. Auto-encoding variational Bayes . In Bengio, Y. and LeCun, Y. (eds.), 2nd International Conference on Learning Representations, ICLR 2014, Banff, AB, Canada, April 14-16, 2014, Conference Track Proceedings , 2014. URL http://arxiv.org/abs/1312.6114

  18. [26]

    G., Shalit, U., and Sontag, D

    Krishnan, R. G., Shalit, U., and Sontag, D. Deep K alman filters. arXiv preprint arXiv:1511.05121, 2015

  19. [27]

    Gradient-based learning applied to document recognition

    LeCun, Y., Bottou, L., Bengio, Y., and Haffner, P. Gradient-based learning applied to document recognition. Proceedings of the IEEE, 86 0 (11): 0 2278--2324, 1998

  20. [28]

    M., Ghahramani, Z., and Valera, I

    Nazabal, A., Olmos, P. M., Ghahramani, Z., and Valera, I. Handling incomplete heterogeneous data using vaes, 2020. URL https://arxiv.org/abs/1807.03653

  21. [29]

    Pytorch: An imperative style, high-performance deep learning library

    Paszke, A., Gross, S., Massa, F., Lerer, A., Bradbury, J., Chanan, G., Killeen, T., Lin, Z., Gimelshein, N., Antiga, L., et al. Pytorch: An imperative style, high-performance deep learning library. In Advances in Neural Information Processing Systems, volume 32, pp.\ 8024--8035, 2019

  22. [30]

    The Gaussian process prior vae for interpretable latent dynamics from pixels

    Pearce, M. The Gaussian process prior vae for interpretable latent dynamics from pixels. In Symposium on Advances in Approximate Bayesian Inference, pp.\ 1--12. PMLR, 2020

  23. [31]

    and Rasmussen, C

    Quinonero-Candela, J. and Rasmussen, C. E. A unifying view of sparse approximate Gaussian process regression. The Journal of Machine Learning Research, 6: 0 1939--1959, 2005

  24. [32]

    a \"a , K., Koskinen, M., and L \

    Ramchandran, S., Tikhonov, G., Kujanp \"a \"a , K., Koskinen, M., and L \"a hdesm \"a ki, H. Longitudinal variational autoencoder. In International Conference on Artificial Intelligence and Statistics, pp.\ 3898--3906. PMLR, 2021

  25. [33]

    Rasmussen, C. E. and Williams, C. K. I. Gaussian Processes for Machine Learning . The MIT Press, 2006. ISBN 026218253X. URL http://www. Gaussian process.org/gpml

  26. [34]

    Generating diverse high-fidelity images with vq-vae-2

    Razavi, A., van den Oord, A., and Vinyals, O. Generating diverse high-fidelity images with vq-vae-2. In Wallach, H., Larochelle, H., Beygelzimer, A., d Alch\' e -Buc, F., Fox, E., and Garnett, R. (eds.), Advances in Neural Information Processing Systems, volume 32. Curran Asso...

  27. [35]

    T., and Duvenaud, D

    Rubanova, Y., Chen, R. T., and Duvenaud, D. K. Latent ordinary differential equations for irregularly-sampled time series. Advances in Neural Information Processing Systems, 32, 2019

  28. [36]

    and Solin, A

    S \"a rkk \"a , S. and Solin, A. Applied stochastic differential equations, volume 10. Cambridge University Press, 2019

  29. [37]

    S \"a rkk \"a , S. et al. Recursive Bayesian inference on stochastic differential equations . Helsinki University of Technology, 2006

  30. [38]

    Solin, A. et al. Stochastic differential equation methods for spatio-temporal Gaussian process regression. PhD Thesis, 2016

  31. [39]

    K., Raiko, T., Maal e, L., S nderby, S

    S nderby, C. K., Raiko, T., Maal e, L., S nderby, S. K., and Winther, O. Ladder variational autoencoders. Advances in Neural Information Processing Systems, 29, 2016

  32. [40]

    T., Klein, A., Falkner, S., and Hutter, F

    Springenberg, J. T., Klein, A., Falkner, S., and Hutter, F. Bayesian optimization with robust Bayesia n neural networks. Advances in Neural Information Processing Systems, 29, 2016

  33. [41]

    L., Chi, Z., and Welty, L

    Stein, M. L., Chi, Z., and Welty, L. J. Approximating likelihoods for large spatial data sets. Journal of the Royal Statistical Society Series B: Statistical Methodology, 66 0 (2): 0 275--296, 2004

  34. [42]

    Variational learning of inducing variables in sparse Gaussian processes

    Titsias, M. Variational learning of inducing variables in sparse Gaussian processes. In International Conference on Artificial Intelligence and Statistics, pp.\ 567--574. PMLR, 2009

  35. [43]

    V., and Filippone, M

    Tran, B.-H., Rossi, S., Milios, D., Michiardi, P., Bonilla, E. V., and Filippone, M. Model selection for B ayesian autoencoders. Advances in Neural Information Processing Systems, 34: 0 19730--19742, 2021 a

  36. [44]

    Fully Bayesian autoencoders with latent sparse Gaussian processes

    Tran, B.-H., Shahbaba, B., Mandt, S., and Filippone, M. Fully Bayesian autoencoders with latent sparse Gaussian processes. In International Conference on Machine Learning, pp.\ 34409--34430. PMLR, 2023

  37. [45]

    Sparse within sparse Gaussian processes using neighbor information

    Tran, G.-L., Milios, D., Michiardi, P., and Filippone, M. Sparse within sparse Gaussian processes using neighbor information. In International Conference on Machine Learning, pp.\ 10369--10378. PMLR, 2021 b

  38. [46]

    Vecchia, A. V. Estimation and model identification for continuous spatial processes. Journal of the Royal Statistical Society Series B: Statistical Methodology, 50 0 (2): 0 297--312, 1988

  39. [47]

    Wu, L., Pleiss, G., and Cunningham, J. P. Variational nearest neighbor Gaussian process. In International Conference on Machine Learning, pp.\ 24114--24130. PMLR, 2022

  40. [48]

    Markovian Gaussian process variational autoencoders

    Zhu, H., Balsells-Rodas, C., and Li, Y. Markovian Gaussian process variational autoencoders. In International Conference on Machine Learning, pp.\ 42938--42961. PMLR, 2023

Pith tools

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