Pith. sign in

REVIEW 4 major objections 5 minor 37 references

How Does the Spatial Distribution of Pre-training Data Affect Geospatial Foundation Models?

T0 review · 4 major / 5 minor · reviewed 2026-08-10 · deepseek-v4-flash

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

Pith's one-line read Balanced and globally representative pre-training data generally outperform region-clustered data for geospatial foundation models in few-shot settings.

desk verdict Useful first systematic comparison of pre-training data sampling for GFMs, but the central 'balanced beats clustered' claim rests mostly on one model-task pair and needs significance testing and tempering. read the letter →

arxiv 2501.12535 v1 pith:2ABI3CE6 submitted 2025-01-21 cs.LG cs.CV

Mirali Purohit , Gedeon Muhawenayo , Esther Rolf , Hannah Kerner This is my paper · ORCID
classification cs.LGcs.CV
keywords geospatialfoundationmodelspre-trainingdatadistributionsamplingstrategiesfew-shotlearningsatelliteimageryEarthobservationrepresentationremotesensing
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 asks whether the geographic distribution of pre-training data changes how well geospatial foundation models perform on downstream Earth-observation tasks. It pre-trains two models — a pixel-timeseries model and a location-encoder model — on five equal-size data compositions ranging from globally balanced to regionally clustered, then fine-tunes each on continent-specific subsets of two globally distributed tasks using only 100 labeled samples per continent. The paper's central finding is that balanced and globally representative compositions (uniform at random, stratified by continent, stratified by biome) generally outperform clustered compositions (intact forests only, or a 50-km radius around the world's largest cities), and that all pretrained variants beat a no-pretraining baseline. It also finds that the performance gaps between compositions shrink as the number of fine-tuning samples grows, so the choice of sampling strategy matters most in the few-shot regime. If correct, the results give GFM developers a concrete, low-cost way to improve model quality: spend the sampling budget on guaranteeing global coverage rather than on simply amassing more data.

What carries the argument

The load-bearing instrument is a controlled resampling pipeline. From a fixed global pool (the ~22 million-sample Presto pool, or Sentinel-2 patches retrieved via a cloud catalog for SatCLIP), the authors create five pre-training compositions of equal size: uniform-at-random over land, stratified by continent, stratified by biome, all within intact-forest cover, and all within 50 km of the world's 10,000 most populated cities. Each composition is used to pre-train one pixel-timeseries GFM and one location-encoder GFM, and each pretrained model is then fine-tuned continent-wise on two global downstream tasks (crop vs. non-crop classification and eco-region classification) with 100 samples per continent, repeated 50 times to average over sampling noise. The pipeline isolates the spatial distribution of pre-training data as the independent variable while holding model architecture, pre-training configuration, pre-training data volume, and fine-tuning procedure fixed.

What would settle it

Compute the land-cover and continental histograms of the ~22-million-sample Presto pool and the overlap between its 'uniform at random' and 'natural forest' subsets; if the two compositions are nearly indistinguishable in their spatial or land-cover statistics, the central comparison collapses. A stronger test is to pre-train each model on pools built from scratch per composition, as the paper did for SatCLIP, and check whether the balanced-vs-clustered ranking persists.

Watch

Extended reading notes

Core claim

The paper claims that, for two structurally different geospatial foundation models, the spatial composition of pre-training data is a first-order factor in downstream few-shot performance. Specifically, all balanced sampling techniques (uniform-at-random, continent-stratified, and biome-stratified) yield approximately equal F1 scores, and these balanced compositions match or exceed clustered compositions (Natural Forest and World Cities) across six continents on both the CropHarvest and EcoRegions tasks. The authors further claim that the relative ranking of compositions is not universal: the city-clustered composition performs on par with balanced ones for the pixel-timeseries model Presto but poorly for the location-encoder model SatCLIP, which they attribute to architectural differences in how each model uses location information.

Load-bearing premise

The comparison assumes that the existing data pools can actually instantiate the intended spatial distributions — in particular, that resampling the ~22-million-sample Presto pool yields compositions as distinct as 'natural forest' and 'uniform at random' are meant to be; if the pool is already skewed geographically or by land cover, the observed differences cannot be attributed to sampling strategy.

Editorial extensions

If this is right

  • GFM pre-training datasets should be curated to guarantee global coverage across continents and biomes rather than clustered by region or environment, at least for few-shot global downstream tasks.
  • Among balanced sampling strategies, the exact method matters little: uniform-at-random, continent-stratified, and biome-stratified perform about equally.
  • Clustered sampling is not categorically harmful; it can match balanced sampling on continents where the cluster naturally dominates (e.g., Natural Forest for crop classification in South America and Oceania).
  • The benefits of any particular sampling strategy fade as fine-tuning data grows, so sampling decisions should be weighted most heavily in low-label deployment regimes.
  • Architecture interacts with data distribution: a location-encoder model is more sensitive to spatially clustered pre-training data than a pixel-timeseries model.

Reading between the lines

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

  • If the generality holds, a practical corollary follows that GFM teams should measure the spatial coverage of their pre-training pool before scaling up data collection, since a biased pool cannot be rescued simply by resampling balanced subsets from it.
  • The architecture-dependence result suggests a testable mechanism: location encoders may internalize the spatial prior directly, so removing location input or adding positional augmentation might reduce the gap between balanced and clustered pre-training.
  • An implicit extension is that deliberately region-specific pre-training could be optimal for region-specific downstream tasks; the paper's continent-wise results provide a template for testing this trade-off at finer granularity than continents.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 5 minor

Summary. The paper studies how the spatial distribution of pre-training data affects the downstream performance of geospatial foundation models. For two GFMs (Presto and SatCLIP), the authors construct five pre-training compositions (uniform random, stratified by continent, stratified by biome, natural-forest-only, and world-cities-only), pre-train each model on equal-sized subsets, and evaluate continent-wise few-shot performance on CropHarvest and EcoRegions with 50 random seeds. The paper claims that balanced, globally representative sampling techniques outperform clustered or region-specific compositions, and that differences diminish as finetuning data grows.

Significance. If established, the paper would provide concrete guidance for GFM pre-training data curation, a topic that is relatively underexplored compared to architecture and pretext-task design. The study is systematic in several respects: it uses two distinct model families, multiple finetuning classifiers, repeated seeds, and a no-pre-training baseline. The authors also transparently acknowledge scope limitations in Appendix E. However, the central claim is only partially supported by the reported results, and the evidence requires either statistical strengthening or a more careful qualification.

major comments (4)
  1. [Section 4, Table 1] The claim that 'all balanced data sampling techniques (i.e., UAR, stratified continent/biome) outperform clustered techniques' is contradicted by the Presto/CropHarvest results. World Cities, the clustered population-centric composition, has no entries highlighted as at least 2% below UAR, and it is numerically above UAR for Africa (0.72 vs 0.71) and North America (0.81 vs 0.80). The only clustered composition that is consistently worse is Natural Forest, and even that equivalence fails for South America and Oceania. The conclusion in Section 5 ('balanced and global representative sampling techniques generally outperform clustered or region-specific compositions') overstates what Table 1 shows for Presto; the assertion needs to be restricted to Natural Forest or to SatCLIP/EcoRegions, or supported with additional evidence that World Cities is meaningfully worse.
  2. [Section 4, Table 1] No significance testing is reported. For the Presto/CropHarvest comparisons, the largest difference between UAR and World Cities on any continent is 0.02, while the standard deviations are 0.02-0.03 across 50 runs. Without paired significance tests (e.g., a paired bootstrap or permutation test on the 50 seeds), the statement that balanced techniques 'usually outperform' clustered techniques is not statistically grounded. Please report confidence intervals for the pairwise differences or explicit significance tests, at least for the headline Table 1 comparisons.
  3. [Section 3.2 and Section 4, Table 1] The EcoRegions task is a 14-class biome classification, and the Stratified Biome pre-training composition is explicitly built by sampling equal numbers per biome using the same Dinerstein et al. (2017) scheme that defines the task labels. This creates a direct alignment between pre-training distribution and downstream label structure, which may explain the large SatCLIP advantage of balanced over clustered compositions. The paper does not acknowledge this potential confound when it generalizes to 'globally diverse pre-training data' being beneficial. The conclusion should be qualified to note that part of the observed effect may be task-specific label alignment rather than generic geographic balance.
  4. [Appendix B, Section 3.3] For Presto, all five compositions are resampled from a single existing pool of approximately 22 million samples (Tseng et al. 2023). The paper does not report the spatial or environmental distribution of this pool. If the pool already has near-global coverage or is biased toward certain regions, the 'Natural Forest' and 'World Cities' subsets may not be as distinct from UAR as intended, weakening the manipulation that the central comparison depends on. Please report the pool's continent/biome distribution, or at least a measure of overlap between the compositions, to verify that the sampling strategies instantiate different intended distributions.
minor comments (5)
  1. [Abstract] There is a typo in the abstract: 'thegeographic' should be 'the geographic'.
  2. [References] The reference for Manas et al. (2021) contains 'uUncurated' which appears to be a typo for 'Uncurated'.
  3. [Appendix C] There are several spelling errors such as 'fineuning' and 'finetuning' used inconsistently; please standardize the spelling.
  4. [Figure 3] The caption for Figure 3 lists all six composition names as subcaptions for every subplot, which is confusing; consider labeling each subplot directly with its composition name.
  5. [Section 4] The sentence 'The reason behind this discrepancy could be the architectural design of the model' is presented without supporting analysis; please mark it explicitly as a hypothesis and, if possible, support it with an analysis of learned representations or a toy experiment.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the central claim is an empirical measurement, and no fitted parameter or self-cited result is used as its own evidence.

full rationale

The paper's central claim is a measured comparison of pre-trained models across five data compositions, evaluated on external downstream tasks (CropHarvest and EcoRegions). No quantity is fitted to the downstream outcomes and then renamed as a prediction; the F1 scores are direct measurements over 50 finetuning seeds. The self-citations to Tseng et al. (2023) and Klemmer et al. (2023) supply data pools, model architectures, and hyperparameter configurations, but those are inputs to the experiment, not the derived conclusion. The claim that balanced compositions outperform clustered ones is not an analytical consequence of how the compositions were defined; it is an empirical result, and the paper's own Table 1 even shows mixed evidence (e.g., World Cities matching UAR on CropHarvest). Concerns about the EcoRegions task labels being biome classes and the stratified-biome pre-training distribution being aligned with those classes are experimental-design or interpretability issues, not circularity. Similarly, the reliance on a pre-existing Presto sample pool is a data-provenance limitation, not a logical reduction of the result to its inputs. No equation in the paper reduces the conclusion to its assumptions by construction.

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

The central claim rests on representative-task assumptions (one downstream task per model, chosen for global coverage), on the validity of continental and biome groupings as diversity axes, on the fidelity of the two existing data pools in realizing the intended sampling distributions, and on the use of shallow finetuning heads as a measure of representation quality. These are domain assumptions, not derived results; the free parameters are experimental design choices (few-shot sample size, equal pre-training counts, and a 2% presentation threshold) rather than fitted values.

free parameters (3)
  • Highlight threshold (2% F1) = 2 percent F1
    Table 1 marks results at least 2% lower than UAR as highlighted; this threshold is hand-set and used for presentation of relative performance, not for statistical inference.
  • Few-shot finetuning sample size = 100
    Section 3.1 fixes n=100 training samples per continent to define the few-shot regime; the value is chosen by the authors and shapes all downstream comparisons.
  • Pre-training sample counts = 6.5M (Presto), 100k (SatCLIP)
    Appendix B equalizes the number of pre-training samples across compositions for each model; the counts are chosen for feasibility and equal strata sizes, not fitted.
assumptions (4)
  • domain assumption One downstream task per GFM (CropHarvest for Presto, EcoRegions for SatCLIP) is representative of the model's general downstream performance.
    Section 3.2 selects tasks based on global coverage and use in the original papers; Appendix E acknowledges the limitation of a single task per model.
  • domain assumption Continental boundaries are a meaningful stratification for evaluating and curating geospatial diversity.
    Section 3.3 uses Esri continents for stratified sampling and continent-wise evaluation; Appendix C explains country-level analysis was infeasible.
  • domain assumption The existing data pools can faithfully realize each intended sampling distribution at equal sample counts.
    Appendix B resamples Presto's 22M pool and builds SatCLIP data from Planetary Computer without reporting the pools' internal spatial coverage.
  • domain assumption Feature extraction plus a shallow finetuning head measures the value of pre-training for downstream tasks.
    Section 3.1 finetunes parametric and non-parametric heads on extracted features, a common GFM evaluation choice that does not capture full fine-tuning behavior.

how reviews work

0 comments
Cite this review

Pith. "Pith review of How Does the Spatial Distribution of Pre-training Data Affect Geospatial Foundation Models?." pith.science (2026). https://pith.science/paper/2ABI3CE6

@misc{pith2026250112535,
  author       = {Pith},
  title        = {Pith review of: How Does the Spatial Distribution of Pre-training Data Affect Geospatial Foundation Models?},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/2ABI3CE6}},
  note         = {Machine review of arXiv:2501.12535}
}
read the original abstract

Foundation models have made rapid advances in many domains including Earth observation, where Geospatial Foundation Models (GFMs) can help address global challenges such as climate change, agriculture, and disaster response. Previous work on GFMs focused on tailoring model architecture and pre-text tasks, and did not investigate the impact of pre-training data selection on model performance. However, recent works from other domains show that the pre-training data distribution is an important factor influencing the performance of the foundation models. With this motivation, our research explores how the geographic distribution of pre-training data affects the performance of GFMs. We evaluated several pre-training data distributions by sampling different compositions from a global data pool. Our experiments with two GFMs on downstream tasks indicate that balanced and globally representative data compositions often outperform region-specific sampling, highlighting the importance of diversity and global coverage in pre-training data. Our results suggest that the most appropriate data sampling technique may depend on the specific GFM architecture. These findings will support the development of robust GFMs by incorporating quality pre-training data distributions, ultimately improving machine learning solutions for Earth observation.

Figures

Figures reproduced from arXiv: 2501.12535 by the authors.

Figure 1
Figure 1. Our experimental pipeline to measure the impact of pre-training data distributions on downstream task performance. [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Performance Comparison across different data [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 4
Figure 4. Continent-wise results for CropHarvest task with Random Forest [PITH_FULL_IMAGE:figures/full_fig_p009_4.png] view at source ↗
Figures from the paper (5 more)
Figure 5
Figure 5. Figure 5: Continent-wise results for CropHarvest task with KNN [PITH_FULL_IMAGE:figures/full_fig_p009_5.png]
Figure 6
Figure 6. Figure 6: Continent-wise results for CropHarvest task with Regression [PITH_FULL_IMAGE:figures/full_fig_p010_6.png]
Figure 7
Figure 7. Figure 7: Continent-wise results for EcoRegions task with Random Forest [PITH_FULL_IMAGE:figures/full_fig_p010_7.png]
Figure 8
Figure 8. Figure 8: Continent-wise results for EcoRegions task with KNN [PITH_FULL_IMAGE:figures/full_fig_p011_8.png]
Figure 9
Figure 9. Figure 9: Continent-wise results for EcoRegions task with MLP [PITH_FULL_IMAGE:figures/full_fig_p011_9.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

37 extracted references · 29 canonical work pages

  1. [1]

    H.; and Baghshah, M

    Abbasi, R.; Rohban, M. H.; and Baghshah, M. S. 2024. Deciphering the Role of Representation Disentanglement: Investigating Compositional Generalization in CLIP Models. In European Conference on Computer Vision. Springer

    work page 2024

  2. [2]

    A.; Das, T.; Pizzati, F.; Torr, P

    Al Kader Hammoud, H. A.; Das, T.; Pizzati, F.; Torr, P. H.; Bibi, A.; and Ghanem, B. 2024. On Pretraining Data Diversity for Self-Supervised Learning. In European Conference on Computer Vision, 54--71. Springer

    work page 2024

  3. [3]

    Christie, G.; Fendley, N.; Wilson, J.; and Mukherjee, R. 2018. Functional Map of the World. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 6172--6180

    work page 2018

  4. [4]

    Cole, E.; Yang, X.; Wilber, K.; Mac Aodha, O.; and Belongie, S. 2022. When Does Contrastive Visual Representation Learning Work? In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 14755--14764

    work page 2022

  5. [5]

    P.; Boudiaf, M.; Culver, D.; Melo, R.; Corro, C.; Martins, A

    Colombo, P.; Pires, T. P.; Boudiaf, M.; Culver, D.; Melo, R.; Corro, C.; Martins, A. F.; Esposito, F.; Raposo, V. L.; Morgado, S.; et al. 2024. Saullm-7b: A Pioneering Large Language Model for Law. arXiv preprint arXiv:2403.03883

    arXiv 2024

  6. [6]

    Cong, Y.; Khanna, S.; Meng, C.; Liu, P.; Rozi, E.; He, Y.; Burke, M.; Lobell, D.; and Ermon, S. 2022. SatMAE : Pre-training Transformers for Temporal and Multi-spectral Satellite Imagery. Advances in Neural Information Processing Systems, 35: 197--211

    work page 2022

  7. [7]

    Dinerstein, E.; Olson, D.; Joshi, A.; Vynne, C.; Burgess, N.; Wikramanayake, E.; Hahn, N.; Palminteri, S.; Hedao, P.; Noss, R.; Hansen, M.; Locke, H.; Ellis, E.; Jones, B.; Barber, C.; Hayes, R.; Kormos, C.; Martin, V.; Crist, E.; and Saleem, M. 2017. An Ecoregion-Based Approach to Protecting Half the Terrestrial Realm. BioScience, 1: 1--12

    work page 2017

  8. [8]

    The Role of Pre-training Data in Transfer Learning

    Entezari, R.; Wortsman, M.; Saukh, O.; Shariatnia, M. M.; Sedghi, H.; and Schmidt, L. 2023. The Role of Pre-training Data in Transfer Learning. arXiv preprint arXiv:2302.13602

    work page Pith review arXiv 2023

Show all 37 references
  1. [9]

    Esri. 2023. World Continents. Accessed January 27, 2024

    2023

  2. [10]

    Fang, A.; Ilharco, G.; Wortsman, M.; Wan, Y.; Shankar, V.; Dave, A.; and Schmidt, L. 2022. Data Determines Distributional Robustness in Contrastive Language Image Pre-training (CLIP). In International Conference on Machine Learning, 6216--6234. PMLR

    2022

  3. [11]

    Fuller, A.; Millard, K.; and Green, J. 2024. CROMA: Remote sensing representations with contrastive radar-optical masked autoencoders. Advances in Neural Information Processing Systems, 36

    2024

  4. [12]

    Hong, D.; Zhang, B.; Li, X.; Li, Y.; Li, C.; Yao, J.; Yokoya, N.; Li, H.; Ghamisi, P.; Jia, X.; et al. 2024. SpectralGPT: Spectral Remote Sensing Foundation Model. IEEE Transactions on Pattern Analysis and Machine Intelligence

    2024

  5. [13]

    Jakubik, J.; Roy, S.; Phillips, C.; Fraccaro, P.; Godwin, D.; Zadrozny, B.; Szwarcman, D.; Gomes, C.; Nyirjesy, G.; Edwards, B.; et al. 2023. Foundation Models for Generalist Geospatial Artificial Intelligence. arXiv preprint arXiv:2310.18660

    2023 arXiv

  6. [14]

    Klemmer, K.; Rolf, E.; Robinson, C.; Mackey, L.; and Ru wurm, M. 2023. SatCLIP: Global, General-Purpose Location Embeddings with Satellite Imagery. arXiv preprint arXiv:2311.17179

    2023 arXiv

  7. [15]

    Longpre, S.; Yauney, G.; Reif, E.; Lee, K.; Roberts, A.; Zoph, B.; Zhou, D.; Wei, J.; Robinson, K.; Mimno, D.; and Ippolito, D. 2024. A Pretrainer ' s Guide to Training Data: Measuring the Effects of Data Age, Domain Coverage, Quality, & Toxicity. In Duh, K.; Gomez, H.; and Be...

    2024

  8. [16]

    Manas, O.; Lacoste, A.; Gir \'o -i Nieto, X.; Vazquez, D.; and Rodriguez, P. 2021. Seasonal Contrast: Unsupervised Pre-training from uUncurated Remote Sensing Data. In Proceedings of the IEEE/CVF International Conference on Computer Vision, 9414--9423

    2021

  9. [17]

    Nedungadi, V.; Kariryaa, A.; Oehmcke, S.; Belongie, S.; Igel, C.; and Lang, N. 2024. MMEarth: Exploring Multi-Modal Pretext Tasks For Geospatial Representation Learning. In European Conference on Computer Vision. Springer

    2024

  10. [18]

    Nguyen, T.; Ilharco, G.; Wortsman, M.; Oh, S.; and Schmidt, L. 2022. Quality Not Quantity: On the Interaction between Dataset Design and Robustness of CLIP. Advances in Neural Information Processing Systems, 35: 21455--21469

    2022

  11. [19]

    Parmar, M.; Mishra, S.; Purohit, M.; Luo, M.; Mohammad, M.; and Baral, C. 2022. In- B o XBART : Get Instructions into Biomedical Multi-Task Learning. In Carpuat, M.; de Marneffe, M.-C.; and Meza Ruiz, I. V., eds., Findings of the Association for Computational Linguistics: NAAC...

    2022

  12. [20]

    Pecher, B.; Srba, I.; and Bielikova, M. 2024. Comparing Specialised Small and General Large Language Models on Text Classification: 100 Labelled Samples to Achieve Break-Even Performance. arXiv preprint arXiv:2402.12819

    2024

  13. [21]

    C.; Laestadius, L.; Turubanova, S.; Yaroshenko, A.; Thies, C.; Smith, W.; Zhuravleva, I.; Komarova, A.; Minnemeyer, S.; et al

    Potapov, P.; Hansen, M. C.; Laestadius, L.; Turubanova, S.; Yaroshenko, A.; Thies, C.; Smith, W.; Zhuravleva, I.; Komarova, A.; Minnemeyer, S.; et al. 2017. The Last Frontiers of Wilderness: Tracking Loss of Intact Forest Landscapes from 2000 to 2013. Science advances, 3(1): e1600821

    2017

  14. [22]

    Ramanujan, V.; Nguyen, T.; Oh, S.; Farhadi, A.; and Schmidt, L. 2024. On the Connection between Pre-training Data Diversity and Fine-tuning Robustness. Advances in Neural Information Processing Systems, 36

    2024

  15. [23]

    J.; Gupta, R.; Li, S.; Brockman, S.; Funk, C.; Clipp, B.; Keutzer, K.; Candido, S.; Uyttendaele, M.; and Darrell, T

    Reed, C. J.; Gupta, R.; Li, S.; Brockman, S.; Funk, C.; Clipp, B.; Keutzer, K.; Candido, S.; Uyttendaele, M.; and Darrell, T. 2023. Scale-MAE : A Scale-aware Masked Autoencoder for Multiscale Geospatial Representation Learning. In Proceedings of the IEEE/CVF International Conf...

    2023

  16. [24]

    Rolf, E.; Klemmer, K.; Robinson, C.; and Kerner, H. 2024. Mission Critical--Satellite Data is a Distinct Modality in Machine Learning. In International Conference on Machine Learning. PMLR

    2024

  17. [25]

    T.; Recht, B.; and Jordan, M

    Rolf, E.; Worledge, T. T.; Recht, B.; and Jordan, M. 2021. Representation matters: Assessing the Importance of Subgroup Allocations in Training Data. In International Conference on Machine Learning, 9040--9051. PMLR

    2021

  18. [26]

    A.; Vakalopoulou, M.; H \"a nsch, R.; Hansen, S.; Nogueira, K.; Prexl, J.; et al

    Roscher, R.; Ru wurm, M.; Gevaert, C.; Kampffmeyer, M.; Dos Santos, J. A.; Vakalopoulou, M.; H \"a nsch, R.; Hansen, S.; Nogueira, K.; Prexl, J.; et al. 2024. Better, Not Just More: Data-Centric Machine Learning for Earth Observation. IEEE Geoscience and Remote Sensing Magazine

    2024

  19. [27]

    Schick, T.; and Sch \"u tze, H. 2021. It ' s Not Just Size That Matters: Small Language Models Are Also Few-Shot Learners. In Toutanova, K.; Rumshisky, A.; Zettlemoyer, L.; Hakkani-Tur, D.; Beltagy, I.; Bethard, S.; Cotterell, R.; Chakraborty, T.; and Zhou, Y., eds., Proceedin...

    2021

  20. [28]

    Shi, J.; Ma, Q.; Ma, H.; and Li, L. 2024. Scaling Law for Time Series Forecasting. arXiv preprint arXiv:2405.15124

    2024 arXiv

  21. [29]

    SimpleMaps. 2024. SimpleMaps World Cities Database. Accessed April 16, 2024

    2024

  22. [30]

    O.; McFarland, M.; Emanuele, R.; Morris, D.; and Augspurger, T

    Source, M. O.; McFarland, M.; Emanuele, R.; Morris, D.; and Augspurger, T. 2022. microsoft/PlanetaryComputer: October 2022

    2022

  23. [31]

    Sumbul, G.; Charfuelan, M.; Demir, B.; and Markl, V. 2019. BigEarthNet: A Large-Scale Benchmark Archive For Remote Sensing Image Understanding. In IGARSS 2019-2019 IEEE International Geoscience and Remote Sensing Symposium, 5901--5904. IEEE

    2019

  24. [32]

    Sun, C.; Shrivastava, A.; Singh, S.; and Gupta, A. 2017. Revisiting Unreasonable Effectiveness of Data in Deep Learning Era. In Proceedings of the IEEE international conference on computer vision, 843--852

    2017

  25. [33]

    Tseng, G.; Cartuyvels, R.; Zvonkov, I.; Purohit, M.; Rolnick, D.; and Kerner, H. 2023. Lightweight, Pre-trained Transformers for Remote Sensing Timeseries. arXiv preprint arXiv:2304.14065

    2023 arXiv

  26. [34]

    L.; and Kerner, H

    Tseng, G.; Zvonkov, I.; Nakalembe, C. L.; and Kerner, H. 2021. Cropharvest: A Global Dataset for Crop-type Classification. In Thirty-fifth Conference on Neural Information Processing Systems Datasets and Benchmarks Track (Round 2)

    2021

  27. [35]

    Wang, Y.; Braham, N. A. A.; Xiong, Z.; Liu, C.; Albrecht, C. M.; and Zhu, X. X. 2023. SSL4EO-S12: A Large-scale Multimodal, Multitemporal Dataset for Self-supervised Learning in Earth Observation [Software and Data Sets]. IEEE Geoscience and Remote Sensing Magazine, 11(3): 98--106

    2023

  28. [36]

    , " * write output.state after.block = add.period write newline

    ENTRY address archivePrefix author booktitle chapter edition editor eid eprint howpublished institution isbn journal key month note number organization pages publisher school series title type volume year label extra.label sort.label short.list INTEGERS output.state before.all...

  29. [37]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 gl...

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.