pith. sign in

arxiv: 2606.29448 · v1 · pith:4KWBPTWGnew · submitted 2026-06-28 · 📊 stat.ME · stat.AP

Scalable Bayesian Spatial Mixture Modelling for Remote Sensing Image Segmentation

Bao Khanh Nguyen , Iain Cameron , Cecilia Balocchi , Torben Sell This is my paper

Pith reviewed 2026-06-30 02:09 UTC · model grok-4.3

classification 📊 stat.ME stat.AP
keywords Bayesian spatial modellingPotts modelremote sensing image segmentationvariational inferenceland cover classificationuncertainty quantificationmixture modelsspatial statistics
0
0 comments X

The pith

Bayesian spatial mixture model enables land cover segmentation in new regions using only external priors

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

The paper proposes a Bayesian extension of the Potts model for segmenting remote sensing images into land cover classes. It incorporates informative priors from labelled images in other regions to avoid needing labels in the target area. This allows the model to quantify uncertainty, account for class interactions, and identify new clusters. The approach uses variational inference for scalability to large images and is demonstrated on Scottish land cover using English data.

Core claim

The central discovery is that the POTTERS framework, by generalizing the spatial dependence structure in the Potts model and using priors estimated from pre-existing labelled data, performs accurate image segmentation on remote sensing data from a target region without any labelled data from that region, while providing robust uncertainty quantification and the ability to detect new clusters.

What carries the argument

The POTTERS model: a Bayesian spatial mixture model extending the Potts model with generalized spatial dependence and informative external priors, approximated by variational inference.

If this is right

  • Classification becomes possible in regions without local labelled data, such as applying England-trained priors to Scotland.
  • Pixel-level uncertainty estimates support more reliable decision-making in conservation monitoring.
  • The model captures interactions between different land cover classes through the spatial mixture.
  • New land cover types not in the prior data can be automatically detected in the target image.
  • Large-scale images can be processed efficiently due to the variational inference algorithm.

Where Pith is reading between the lines

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

  • If the prior transfer works across moderate shifts, the method could reduce reliance on costly ground-truth surveys worldwide.
  • Applying the same framework to multi-temporal data might allow tracking land cover changes over time without new labels.
  • Testing on regions with larger distribution shifts, such as different continents, would reveal the limits of prior transfer.
  • The variational approximation could be compared to MCMC for accuracy on smaller images to validate scalability claims.

Load-bearing premise

Priors estimated from labelled data in one region, such as England, transfer effectively to a different target region like Scotland despite possible distribution shifts in the image data.

What would settle it

A substantial drop in segmentation accuracy or failure to detect known classes when the target region's land cover distribution differs markedly from the source regions used for priors.

Figures

Figures reproduced from arXiv: 2606.29448 by Bao Khanh Nguyen, Cecilia Balocchi, Iain Cameron, Torben Sell.

Figure 1
Figure 1. Figure 1: Simulated data sets. For each dataset, the left panel shows a heatmap of the [PITH_FULL_IMAGE:figures/full_fig_p014_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: The result maps produced by POTTERS in two situations: (a) without the Potts [PITH_FULL_IMAGE:figures/full_fig_p015_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: (a) Glensaugh region satellite image, March–May 2024. (b) England region [PITH_FULL_IMAGE:figures/full_fig_p017_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Maps obtained from applying the POTTERS model to the Glensaugh area. [PITH_FULL_IMAGE:figures/full_fig_p018_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Comparison between the predicted classes and the ground truth labels. [PITH_FULL_IMAGE:figures/full_fig_p019_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: The satellite images of the target region near Glensaugh (Scotland) corresponding [PITH_FULL_IMAGE:figures/full_fig_p038_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: The RGB satellite image of the North-Eastern region in England as external data [PITH_FULL_IMAGE:figures/full_fig_p038_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Probability maps for each cluster. 40 [PITH_FULL_IMAGE:figures/full_fig_p040_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Accuracy map 41 [PITH_FULL_IMAGE:figures/full_fig_p041_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: Confusion matrix 42 [PITH_FULL_IMAGE:figures/full_fig_p042_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: Maps for secondary-level classes 43 [PITH_FULL_IMAGE:figures/full_fig_p043_11.png] view at source ↗
read the original abstract

Accurate and scalable land cover classification is essential for global conservation monitoring and policy-making. While remote sensing images provide a cost-effective alternative to ground surveys, current methods often lack principled uncertainty quantification and require substantial labelled data, limiting their usability and reliability in new regions with distribution shifts. We propose a Bayesian spatial mixture modelling approach for image segmentation, extending the classical Potts model by allowing for a generalised spatial dependence structure and incorporating informative priors estimated from pre-existing labelled data. Our framework, called POTTERS (Potts Model for Enhanced Remote Sensing), enables robust uncertainty quantification, accounts for class interactions, and can detect new clusters in the target region of interest. Crucially, our model does not require labelled data from the target region; instead, it incorporates prior information about the labels from pre-existing externally labelled images. To ensure scalability to large remote sensing images, we develop an efficient variational inference algorithm for posterior approximation. We demonstrate the benefits of our approach in simulation studies and apply it to land cover classification in a case study in Scotland, leveraging publicly available remote sensing data from England.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The paper proposes POTTERS, an extension of the classical Potts model for Bayesian spatial mixture modeling of remote sensing images. It incorporates informative priors on labels and class interactions estimated from externally labeled source images (e.g., England) to enable segmentation, robust uncertainty quantification, class-interaction modeling, and detection of new clusters in an unlabeled target region (e.g., Scotland), without requiring any labeled target data. Scalability is achieved via a custom variational inference algorithm. The approach is illustrated on simulation studies and a Scotland land-cover case study using publicly available remote-sensing data.

Significance. If the external priors transfer reliably and the variational scheme yields accurate posterior approximations, the framework would offer a practical route to uncertainty-aware land-cover mapping in new regions with minimal labeling cost. The explicit handling of spatial dependence and new-cluster detection distinguishes it from standard mixture models, and the use of external data for priors is a concrete strength when the transfer assumption holds.

major comments (2)
  1. [Case study] Scotland case study: no diagnostic, sensitivity analysis, or comparison against an uninformative-prior or target-only baseline is reported for the transfer of England-derived label marginals and interaction parameters to Scotland. This is load-bearing for the central claim that the model requires no labeled target data, because any substantial distribution shift would propagate directly into the reported posterior uncertainty and new-cluster detections.
  2. [Variational inference] Variational inference section: the manuscript provides no quantitative assessment (e.g., simulation recovery of known posteriors or comparison to MCMC on moderate-sized images) of how accurately the variational approximation recovers the true posterior, especially the uncertainty quantification and new-cluster detection components. This directly affects the reliability of the “robust uncertainty quantification” claim.
minor comments (2)
  1. [Model definition] Notation for the generalized spatial dependence structure is introduced without an explicit comparison table to the standard Potts model, making it harder to see exactly which parameters are new.
  2. [Figures] Figure captions for the Scotland results should state the number of pixels, the number of classes, and whether any England data were held out for validation.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their constructive comments, which highlight important aspects of our claims regarding prior transfer and variational approximation quality. We address each point below and will revise the manuscript to incorporate additional analyses.

read point-by-point responses

  1. Referee: [Case study] Scotland case study: no diagnostic, sensitivity analysis, or comparison against an uninformative-prior or target-only baseline is reported for the transfer of England-derived label marginals and interaction parameters to Scotland. This is load-bearing for the central claim that the model requires no labeled target data, because any substantial distribution shift would propagate directly into the reported posterior uncertainty and new-cluster detections.

    Authors: We agree that explicit diagnostics on prior transfer are needed to support the central claim. In the revised manuscript we will add a sensitivity analysis that varies the strength of the England-derived priors and a direct comparison to an uninformative-prior baseline fitted to the Scotland data alone; these will quantify changes in posterior uncertainty and new-cluster detections. revision: yes

  2. Referee: [Variational inference] Variational inference section: the manuscript provides no quantitative assessment (e.g., simulation recovery of known posteriors or comparison to MCMC on moderate-sized images) of how accurately the variational approximation recovers the true posterior, especially the uncertainty quantification and new-cluster detection components. This directly affects the reliability of the “robust uncertainty quantification” claim.

    Authors: The existing simulation studies demonstrate parameter recovery, but we acknowledge the absence of direct variational-vs-MCMC comparisons. We will expand the simulation section to include quantitative metrics (e.g., posterior mean and variance recovery, cluster detection accuracy) on moderate-sized images where MCMC is feasible, thereby strengthening the uncertainty-quantification claims. revision: yes

Circularity Check

0 steps flagged

No significant circularity; derivation uses external priors and independent variational inference

full rationale

The paper's central claims rest on a Bayesian extension of the Potts model that incorporates priors estimated from separate external labelled images (England data) and applies variational inference for posterior approximation on the target region (Scotland). No equations or steps in the provided abstract or description reduce a prediction or result to a fitted input by construction, nor does any load-bearing premise rely on self-citation chains. The model development, scalability algorithm, and case-study application are presented as independent methodological contributions supported by simulation studies and external data transfer, making the derivation self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Abstract-only review provides no explicit free parameters, axioms, or invented entities; the approach relies on standard Bayesian modeling assumptions and the Potts model extension whose details are not provided.

pith-pipeline@v0.9.1-grok · 5717 in / 1120 out tokens · 33683 ms · 2026-06-30T02:09:56.503350+00:00 · methodology

discussion (0)

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

Reference graph

Works this paper leans on

190 extracted references · 84 canonical work pages · 8 internal anchors

  1. [1]

    , month = mar, year =

    Robinson, James C. , month = mar, year =. An. Cambridge Aspire website , publisher =. doi:10.1017/9781139030267 , note =

    work page doi:10.1017/9781139030267

  2. [2]

    Bayesian nonparametric priors for hidden

    Lü, Hongliang and Arbel, Julyan and Forbes, Florence , month = jul, year =. Bayesian nonparametric priors for hidden. Statistics and Computing , publisher =. doi:10.1007/s11222-020-09935-9 , abstract =

    work page doi:10.1007/s11222-020-09935-9

  3. [3]

    Department of Statistical Science, Duke University, Durham, NC , author =

    Bayesian. Department of Statistical Science, Duke University, Durham, NC , author =

  4. [4]

    Besag, Julian , year =. On the. Journal of the Royal Statistical Society. Series B (Methodological) , publisher =

  5. [5]

    Vaart, A. W. van der , year =. Asymptotic. doi:10.1017/CBO9780511802256 , abstract =

    work page doi:10.1017/cbo9780511802256

  6. [6]

    and van der Vaart, Aad W

    Ghosal, Subhashis and Ghosh, Jayanta K. and van der Vaart, Aad W. , year =. Convergence. The Annals of Statistics , publisher =

  7. [7]

    Posterior

    Lee, Kyeongwon and Lin, Lizhen and Park, Jaewoo and Jeong, Seonghyun , month = oct, year =. Posterior. doi:10.48550/arXiv.2506.19144 , abstract =

    work page doi:10.48550/arxiv.2506.19144

  8. [8]

    Nonparametric regression using deep neural networks with ReLU activation function

    Nonparametric regression using deep neural networks with. The Annals of Statistics , author =. doi:10.1214/19-AOS1875 , language =

    work page doi:10.1214/19-aos1875

  9. [9]

    Bernoulli , author =

    Rates of contraction of posterior distributions based on p-exponential priors , volume =. Bernoulli , author =. doi:10.3150/20-BEJ1285 , language =

    work page doi:10.3150/20-bej1285

  10. [10]

    Bayesian neural network priors for edge-preserving inversion , url =

    Li, Chen and Dunlop, Matthew and Stadler, Georg , month = dec, year =. Bayesian neural network priors for edge-preserving inversion , url =. doi:10.48550/arXiv.2112.10663 , abstract =

    work page doi:10.48550/arxiv.2112.10663

  11. [11]

    and Briol, François-Xavier , month = jan, year =

    Matsubara, Takuo and Oates, Chris J. and Briol, François-Xavier , month = jan, year =. The. doi:10.48550/arXiv.2010.08488 , abstract =

    work page doi:10.48550/arxiv.2010.08488 2010

  12. [12]

    Atanov, Andrei and Ashukha, Arsenii and Struminsky, Kirill and Vetrov, Dmitry and Welling, Max , month = nov, year =. The. doi:10.48550/arXiv.1810.06943 , abstract =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.1810.06943

  13. [13]

    and Wenzel, Florian and Rätsch, Gunnar and Turner, Richard E

    Fortuin, Vincent and Garriga-Alonso, Adrià and Ober, Sebastian W. and Wenzel, Florian and Rätsch, Gunnar and Turner, Richard E. and Wilk, Mark van der and Aitchison, Laurence , month = mar, year =. Bayesian. doi:10.48550/arXiv.2102.06571 , abstract =

    work page doi:10.48550/arxiv.2102.06571

  14. [14]

    Conjugate

    Laplante, William and Altamirano, Matias and Knoblauch, Jeremias and Duncan, Andrew and Briol, François-Xavier , month = nov, year =. Conjugate. doi:10.48550/arXiv.2511.23275 , abstract =

    work page doi:10.48550/arxiv.2511.23275

  15. [16]

    Statistics and Computing , author =

    Model-based clustering based on sparse finite. Statistics and Computing , author =. 2016 , note =. doi:10.1007/s11222-014-9500-2 , abstract =

    work page doi:10.1007/s11222-014-9500-2 2016

  16. [17]

    and Roy, Hrittik and Krämer, Nicholas and Zainchkovskyy, Yevgen and Syrota, Stas and Mahou, Alejandro Valverde and Ek, Carl Henrik and Hauberg, Søren , month = oct, year =

    Fadel, Samuel G. and Roy, Hrittik and Krämer, Nicholas and Zainchkovskyy, Yevgen and Syrota, Stas and Mahou, Alejandro Valverde and Ek, Carl Henrik and Hauberg, Søren , month = oct, year =. doi:10.48550/arXiv.2510.23684 , abstract =

    work page doi:10.48550/arxiv.2510.23684

  17. [18]

    Bayesian neural networks with interpretable priors from

    Alberts, Alex and Bilionis, Ilias , month = oct, year =. Bayesian neural networks with interpretable priors from. doi:10.48550/arXiv.2510.23745 , abstract =

    work page doi:10.48550/arxiv.2510.23745

  18. [19]

    Baragilly, Mohammed and Gabr, Hend , month = oct, year =. High-. doi:10.48550/arXiv.2510.14145 , abstract =

    work page doi:10.48550/arxiv.2510.14145

  19. [20]

    Uncertainty

    Wilson, Joseph and Heide, Chris van der and Hodgkinson, Liam and Roosta, Fred , month = oct, year =. Uncertainty. doi:10.48550/arXiv.2502.02870 , abstract =

    work page doi:10.48550/arxiv.2502.02870

  20. [21]

    , month = apr, year =

    Ma, Tianyi and Wang, Tengyao and Samworth, Richard J. , month = apr, year =. Deep learning with missing data , url =. doi:10.48550/arXiv.2504.15388 , abstract =

    work page doi:10.48550/arxiv.2504.15388

  21. [22]

    and Ročková, Veronika , month = oct, year =

    Jeong, So Won and Zhai, Percy S. and Ročková, Veronika , month = oct, year =. Conditional. doi:10.48550/arXiv.2510.09534 , abstract =

    work page doi:10.48550/arxiv.2510.09534

  22. [23]

    Leveraging

    Yang, Xiliang and Xiong, Yifei and He, Zhijian , month = oct, year =. Leveraging. doi:10.48550/arXiv.2401.16776 , abstract =

    work page doi:10.48550/arxiv.2401.16776

  23. [24]

    Boothe and Maroulas, Vasileios , month = oct, year =

    Dayton, Sarah Harkins and Everett, Hayden and Schizas, Ioannis and Jr, David L. Boothe and Maroulas, Vasileios , month = oct, year =. Bayesian. doi:10.48550/arXiv.2510.11704 , abstract =

    work page doi:10.48550/arxiv.2510.11704

  24. [25]

    Biostatistics , author =

    A. Biostatistics , author =. 2019 , pages =. doi:10.1093/biostatistics/kxy019 , abstract =

    work page doi:10.1093/biostatistics/kxy019 2019

  25. [26]

    New region force for variational models in image segmentation and high dimensional data clustering

    Wei, Ke and Yin, Ke and Tai, Xue-Cheng and Chan, Tony F. , month = apr, year =. New region force for variational models in image segmentation and high dimensional data clustering , url =. doi:10.48550/arXiv.1704.08218 , abstract =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.1704.08218

  26. [27]

    A double

    Liang, Faming , month = sep, year =. A double. Journal of Statistical Computation and Simulation , publisher =. doi:10.1080/00949650902882162 , abstract =

    work page doi:10.1080/00949650902882162

  27. [28]

    Functional Variational Bayesian Neural Networks

    Sun, Shengyang and Zhang, Guodong and Shi, Jiaxin and Grosse, Roger , month = mar, year =. Functional. doi:10.48550/arXiv.1903.05779 , abstract =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.1903.05779 1903

  28. [29]

    and Byun, Yewon and Kolter, J

    Sam, Dylan and Pukdee, Rattana and Jeong, Daniel P. and Byun, Yewon and Kolter, J. Zico , month = feb, year =. Bayesian. doi:10.48550/arXiv.2402.13410 , abstract =

    work page doi:10.48550/arxiv.2402.13410

  29. [30]

    Fast predictive uncertainty for classification with

    Hobbhahn, Marius and Kristiadi, Agustinus and Hennig, Philipp , month = aug, year =. Fast predictive uncertainty for classification with. Proceedings of the

  30. [31]

    D'Angelo, Francesco and Fortuin, Vincent and Wenzel, Florian , month = jun, year =. On. doi:10.48550/arXiv.2106.10760 , abstract =

    work page doi:10.48550/arxiv.2106.10760

  31. [32]

    Neural Networks , author =

    Bayesian approach for neural networks—review and case studies , volume =. Neural Networks , author =. 2001 , pages =. doi:10.1016/S0893-6080(00)00098-8 , abstract =

    work page doi:10.1016/s0893-6080(00)00098-8 2001

  32. [33]

    and Rudner, Tim G

    Papamarkou, Theodore and Skoularidou, Maria and Palla, Konstantina and Aitchison, Laurence and Arbel, Julyan and Dunson, David and Filippone, Maurizio and Fortuin, Vincent and Hennig, Philipp and Hernández-Lobato, José Miguel and Hubin, Aliaksandr and Immer, Alexander and Karaletsos, Theofanis and Khan, Mohammad Emtiyaz and Kristiadi, Agustinus and Li, Yi...

    work page doi:10.48550/arxiv.2402.00809

  33. [34]

    International Statistical Review , author =

    Priors in. International Statistical Review , author =. 2022 , pages =. doi:10.1111/insr.12502 , abstract =

    work page doi:10.1111/insr.12502 2022

  34. [35]

    Neural Computation , author =

    Data. Neural Computation , author =. 1997 , pages =. doi:10.1162/neco.1997.9.8.1805 , abstract =

    work page doi:10.1162/neco.1997.9.8.1805 1997

  35. [36]

    Scanvic, Jérémy and Davies, Mike and Abry, Patrice and Tachella, Julián , month = mar, year =. Scale-. doi:10.48550/arXiv.2312.11232 , abstract =

    work page doi:10.48550/arxiv.2312.11232

  36. [37]

    Cambridge Explorations in Arts and Sciences , author =

    Auto-. Cambridge Explorations in Arts and Sciences , author =. doi:10.61603/ceas.v2i1.33 , abstract =

    work page doi:10.61603/ceas.v2i1.33

  37. [38]

    Auto-Encoding Variational Bayes

    Kingma, Diederik P. and Welling, Max , month = dec, year =. Auto-. doi:10.48550/arXiv.1312.6114 , abstract =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.1312.6114

  38. [39]

    Prakash, Mangal and Krull, Alexander and Jug, Florian , month = mar, year =. Fully. doi:10.48550/arXiv.2006.06072 , abstract =

    work page doi:10.48550/arxiv.2006.06072 2006

  39. [40]

    Entropy , author =

    Self-. Entropy , author =. 2021 , note =. doi:10.3390/e23060747 , abstract =

    work page doi:10.3390/e23060747 2021

  40. [41]

    Journal of the Royal Statistical Society Series B: Statistical Methodology , author =

    Martingale posterior distributions , volume =. Journal of the Royal Statistical Society Series B: Statistical Methodology , author =. 2024 , pages =. doi:10.1093/jrsssb/qkad005 , abstract =

    work page doi:10.1093/jrsssb/qkad005 2024

  41. [42]

    Journal of Computational and Graphical Statistics , author =

    Martingale. Journal of Computational and Graphical Statistics , author =. 2025 , pages =. doi:10.1080/10618600.2024.2441997 , abstract =

    work page doi:10.1080/10618600.2024.2441997 2025

  42. [43]

    Journal of the Royal Statistical Society Series B: Statistical Methodology , author =

    Trace-class. Journal of the Royal Statistical Society Series B: Statistical Methodology , author =. 2023 , pages =. doi:10.1093/jrsssb/qkac005 , abstract =

    work page doi:10.1093/jrsssb/qkac005 2023

  43. [44]

    IEEE Transactions on Knowledge and Data Engineering , author =

    Towards. IEEE Transactions on Knowledge and Data Engineering , author =. 2016 , pages =. doi:10.1109/TKDE.2016.2606428 , abstract =

    work page doi:10.1109/tkde.2016.2606428 2016

  44. [45]

    Tschandl, Philipp and. The. 2018 , note =. doi:10.7910/DVN/DBW86T , abstract =

    work page doi:10.7910/dvn/dbw86t 2018

  45. [46]

    IEEE Transactions on Biomedical Engineering , author =

    Supervised. IEEE Transactions on Biomedical Engineering , author =. 2015 , note =. doi:10.1109/TBME.2014.2344660 , abstract =

    work page doi:10.1109/tbme.2014.2344660 2015

  46. [47]

    IEEE Transactions on Signal Processing , author =

    Sequential. IEEE Transactions on Signal Processing , author =. 2023 , keywords =. doi:10.1109/TSP.2023.3267992 , abstract =

    work page doi:10.1109/tsp.2023.3267992 2023

  47. [48]

    International Journal of Computer Vision , author =

    Nonparametric. International Journal of Computer Vision , author =. 2008 , keywords =. doi:10.1007/s11263-007-0061-0 , abstract =

    work page doi:10.1007/s11263-007-0061-0 2008

  48. [49]

    Statistics and Computing , author =

    Variational. Statistics and Computing , author =. 2009 , keywords =. doi:10.1007/s11222-008-9095-6 , abstract =

    work page doi:10.1007/s11222-008-9095-6 2009

  49. [50]

    Rydén, Tobias and Titterington, D. M. , month = jun, year =. Computational. Journal of Computational and Graphical Statistics , publisher =. doi:10.1080/10618600.1998.10474770 , abstract =

    work page doi:10.1080/10618600.1998.10474770 1998

  50. [51]

    Scalable Bayesian Inference for the Inverse Temperature of a Hidden Potts Model

    Moores, Matthew T. and Nicholls, Geoff K. and Pettitt, Anthony N. and Mengersen, Kerrie , month = aug, year =. Scalable. doi:10.48550/arXiv.1503.08066 , abstract =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.1503.08066

  51. [52]

    Li, Stan Z , year =. Markov

  52. [53]

    Variational Bayesian Approach and Gauss-Markov-Potts prior model

    Chapdelaine, Camille , month = aug, year =. Variational. doi:10.48550/arXiv.1808.09552 , abstract =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.1808.09552

  53. [54]

    Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences , author =

    Bayesian cluster analysis , volume =. Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences , author =. 2023 , keywords =. doi:10.1098/rsta.2022.0149 , abstract =

    work page doi:10.1098/rsta.2022.0149 2023

  54. [55]

    Li, Qianyun and Yi, Faliu and Wang, Tao and Xiao, Guanghua and Liang, Faming , month = jan, year =. Lung. Cancer Informatics , publisher =. doi:10.1177/1176935117711910 , abstract =

    work page doi:10.1177/1176935117711910

  55. [56]

    and Farag, Aly A

    Ali, Asem M. and Farag, Aly A. and Gimel'farb, Georgy , month = dec, year =. Analytical method for. 2008 19th. doi:10.1109/ICPR.2008.4761339 , abstract =

    work page doi:10.1109/icpr.2008.4761339 2008

  56. [57]

    Physical Review Research , author =

    Mapping of attention mechanisms to a generalized. Physical Review Research , author =. 2024 , pages =. doi:10.1103/PhysRevResearch.6.023057 , abstract =

    work page doi:10.1103/physrevresearch.6.023057 2024

  57. [58]

    Yin, Mingzhang and Wang, Y. X. Rachel and Sarkar, Purnamrita , month = mar, year =. A. doi:10.48550/arXiv.1907.12203 , abstract =

    work page doi:10.48550/arxiv.1907.12203 1907

  58. [59]

    Structured

    Alexos, Antonios and Boyd, Alex and Mandt, Stephan , month = jul, year =. Structured. doi:10.48550/arXiv.2107.09028 , abstract =

    work page doi:10.48550/arxiv.2107.09028

  59. [60]

    Laurie , year =

    Kindermann, Ross and Snell, J. Laurie , year =. Markov. American Mathematical Society , publisher =. doi:10.1090/conm/001 , abstract =

    work page doi:10.1090/conm/001

  60. [61]

    Computational Statistics & Data Analysis , author =

    Mean-field variational approximate. Computational Statistics & Data Analysis , author =. 2007 , keywords =. doi:10.1016/j.csda.2006.10.028 , abstract =

    work page doi:10.1016/j.csda.2006.10.028 2007

  61. [62]

    Beal, Matthew J and Ghahramani, Zoubin , editor =. The. Bayesian. 2003 , pages =. doi:10.1093/oso/9780198526155.003.0025 , abstract =

    work page doi:10.1093/oso/9780198526155.003.0025 2003

  62. [63]

    IEEE Transactions on Systems, Man, and Cybernetics, Part B (Cybernetics) , author =

    Variational learning for. IEEE Transactions on Systems, Man, and Cybernetics, Part B (Cybernetics) , author =. 2006 , note =. doi:10.1109/TSMCB.2006.872273 , abstract =

    work page doi:10.1109/tsmcb.2006.872273 2006

  63. [64]

    A practical tutorial on

    Tran, Minh-Ngoc and Nguyen, Trong-Nghia and Dao, Viet-Hung , month = mar, year =. A practical tutorial on. doi:10.48550/arXiv.2103.01327 , abstract =

    work page doi:10.48550/arxiv.2103.01327

  64. [65]

    Artificial Intelligence Review , author =

    A tutorial on variational. Artificial Intelligence Review , author =. 2012 , keywords =. doi:10.1007/s10462-011-9236-8 , abstract =

    work page doi:10.1007/s10462-011-9236-8 2012

  65. [66]

    Computer Vision and Image Understanding , author =

    Variational inference for medical image segmentation , volume =. Computer Vision and Image Understanding , author =. 2016 , keywords =. doi:10.1016/j.cviu.2016.04.004 , abstract =

    work page doi:10.1016/j.cviu.2016.04.004 2016

  66. [67]

    Tibshirani, and Larry Wasserman

    Variational. Journal of the American Statistical Association , author =. 2017 , pages =. doi:10.1080/01621459.2017.1285773 , abstract =

    work page doi:10.1080/01621459.2017.1285773 2017

  67. [68]

    and Kucukelbir, Alp and McAuliffe, Jon D

    Blei, David M. and Kucukelbir, Alp and McAuliffe, Jon D. , month = apr, year =. Variational. Journal of the American Statistical Association , publisher =

  68. [69]

    Dilokthanakul, Nat and Mediano, Pedro A. M. and Garnelo, Marta and Lee, Matthew C. H. and Salimbeni, Hugh and Arulkumaran, Kai and Shanahan, Murray , month = jan, year =. Deep

  69. [70]

    Wu, F. Y. , month = jan, year =. The. Reviews of Modern Physics , publisher =. doi:10.1103/RevModPhys.54.235 , abstract =

    work page doi:10.1103/revmodphys.54.235

  70. [71]

    , month = dec, year =

    Cipra, Barry A. , month = dec, year =. An. The American Mathematical Monthly , publisher =. doi:10.2307/2322600 , abstract =

    work page doi:10.2307/2322600

  71. [72]

    Variational

    Jiang, Zhuxi and Zheng, Yin and Tan, Huachun and Tang, Bangsheng and Zhou, Hanning , month = jun, year =. Variational

  72. [73]

    Goyal, R

    Prasad, Vignesh and Das, Dipanjan and Bhowmick, Brojeshwar , month = jul, year =. Variational. 2020. doi:10.1109/IJCNN48605.2020.9207523 , abstract =

    work page doi:10.1109/ijcnn48605.2020.9207523 2020

  73. [74]

    Goyal, R

    Li, Zhihan and Zhao, Youjian and Xu, Haowen and Chen, Wenxiao and Xu, Shangqing and Li, Yilin and Pei, Dan , month = jul, year =. Unsupervised. 2020. doi:10.1109/IJCNN48605.2020.9207493 , abstract =

    work page doi:10.1109/ijcnn48605.2020.9207493 2020

  74. [75]

    NeuroImage , author =

    Variational. NeuroImage , author =. 2013 , keywords =. doi:10.1016/j.neuroimage.2013.03.008 , abstract =

    work page doi:10.1016/j.neuroimage.2013.03.008 2013

  75. [76]

    Pottics –

    Dann, Christoph and Gehler, Peter and Roth, Stefan and Nowozin, Sebastian , editor =. Pottics –. Pattern. 2012 , note =. doi:10.1007/978-3-642-32717-9_40 , abstract =

    work page doi:10.1007/978-3-642-32717-9_40 2012

  76. [77]

    , month = jan, year =

    Zobay, O. , month = jan, year =. Variational. Electronic Journal of Statistics , publisher =. doi:10.1214/14-EJS887 , abstract =

    work page doi:10.1214/14-ejs887

  77. [78]

    Biometrika , author =

    Bayesian cluster analysis , volume =. Biometrika , author =. 1978 , pages =. doi:10.1093/biomet/65.1.31 , abstract =

    work page doi:10.1093/biomet/65.1.31 1978

  78. [79]

    Auto-encoder

    Song, Chunfeng and Liu, Feng and Huang, Yongzhen and Wang, Liang and Tan, Tieniu , editor =. Auto-encoder. Progress in. 2013 , pages =

    2013

  79. [80]

    Variational

    Guan, Yue and Dy, Jennifer G and Niu, Donglin and Ghahramani, Zoubin , year =. Variational

  80. [81]

    doi:10.1007/978-3-642-41822-8 , language =

    Progress in. doi:10.1007/978-3-642-41822-8 , language =

    work page doi:10.1007/978-3-642-41822-8

Showing first 80 references.