pith. sign in

arxiv: 2605.31068 · v1 · pith:JG2GCK5Pnew · submitted 2026-05-29 · 💻 cs.CV

HQ-JEPA: Hybrid Quantum Joint-Embedding Predictive Architecture for Cross-Modal Remote Sensing Representation Learning

Md Aminur Hossain , Ayush V. Patel , Sanjay K. Singh , Biplab Banerjee This is my paper

Pith reviewed 2026-06-28 23:17 UTC · model grok-4.3

classification 💻 cs.CV
keywords remote sensingself-supervised learningcross-modal learningrepresentation learningquantum machine learningJEPASentinel imageryGeoBench
0
0 comments X

The pith

HQ-JEPA adds a quantum fidelity loss to JEPA-style masked prediction to align features from paired Sentinel radar and optical images.

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

The paper sets out to establish that a hybrid quantum-classical joint-embedding predictive architecture can produce stronger semantic representations for remote sensing by extending masked latent prediction across modalities while adding quantum state-overlap regularization. If correct, this would let models learn aligned features from heterogeneous sensors without pixel reconstruction and would improve transfer to classification and segmentation. A sympathetic reader cares because remote sensing applications routinely combine optical and radar data yet still lack robust self-supervised methods that exploit both predictive and cross-modal signals. The evaluation on GeoBench under linear probing and fine-tuning is presented as evidence that the combined objectives outperform or match strong classical baselines.

Core claim

HQ-JEPA extends JEPA-style masked latent prediction to paired Sentinel-1 and Sentinel-2 imagery by predicting masked target representations from visible context regions while aligning heterogeneous modality features in a shared embedding space. Four objectives are combined: latent token prediction, cross-modal token alignment, SIGReg-based Gaussian regularization, and a differentiable SWAP-test-based Fidelity Quantum Similarity loss. The resulting encoder, when evaluated on GeoBench classification and segmentation tasks, achieves competitive and often superior performance over strong self-supervised and remote sensing foundation-model baselines.

What carries the argument

The differentiable SWAP-test-based Fidelity Quantum Similarity (FQS) loss, which supplies quantum state-overlap similarity as an additional regularization signal in the fused latent space.

If this is right

  • Masked latent prediction in a shared embedding space produces semantic representations without requiring pixel-level reconstruction of Sentinel imagery.
  • Cross-modal token alignment regularizes the latent space so that radar and optical features become comparable for downstream tasks.
  • Gaussian regularization in the fused space further stabilizes the joint embedding learned from paired modalities.
  • The pretrained encoder transfers to both linear probing and full fine-tuning on classification and segmentation benchmarks.

Where Pith is reading between the lines

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

  • The same combination of predictive and quantum regularization objectives could be tested on other paired sensor types such as hyperspectral and LiDAR data.
  • If the SWAP-test simulation scales efficiently, the FQS term might be replaced by a true quantum circuit on future hardware while keeping the rest of the pipeline unchanged.
  • The framework implies that quantum-inspired similarity measures can act as drop-in regularizers even when run on classical simulators.

Load-bearing premise

The Fidelity Quantum Similarity loss must supply an independent and beneficial regularization signal that improves representations beyond what the classical prediction and alignment objectives already achieve.

What would settle it

An ablation experiment in which removing the FQS loss from the training objectives leaves performance on the GeoBench linear-probing and fine-tuning tasks unchanged or lower than the full model.

Figures

Figures reproduced from arXiv: 2605.31068 by Ayush V. Patel, Biplab Banerjee, Md Aminur Hossain, Sanjay K. Singh.

Figure 1
Figure 1. Figure 1: Overview of HQ-JEPA. The role of representation geometry in the generalization and stability of machine learn￾ing models has been widely studied in recent years [17]. The results show that structured em￾bedding distributions improve transfer perfor￾mance and prevent collapse in non-contrastive self-supervised learning [47]. However, the ef￾fects of distributional regularization on cross￾modal predictive le… view at source ↗
Figure 2
Figure 2. Figure 2: HQ-JEPA architecture for multi-modal remote sensing: S2 and S1 patches undergo [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: a) Hybrid Mamba-ViT encoder integrating patch/positional embeddings with [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Illustration of the adopted I-JEPA masking strategy showing the original image, [PITH_FULL_IMAGE:figures/full_fig_p007_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Quantum circuit for SWAP test between predicted student and teacher states: stu [PITH_FULL_IMAGE:figures/full_fig_p009_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Training dynamics of HQ-JEPA over 400 epochs, showing the evolution of the [PITH_FULL_IMAGE:figures/full_fig_p011_6.png] view at source ↗
read the original abstract

We introduce HQ-JEPA, a hybrid quantum-classical joint-embedding predictive architecture for cross-modal remote sensing representation learning. The proposed framework extends JEPA-style masked latent prediction to paired Sentinel-1 and Sentinel-2 imagery by predicting masked target representations from visible context regions while aligning heterogeneous modality features in a shared embedding space. To improve representation quality, HQ-JEPA combines four complementary objectives: latent token prediction, cross-modal token alignment, SIGReg-based Gaussian regularization in the fused latent space, and a differentiable SWAP-test-based Fidelity Quantum Similarity (FQS) loss. Unlike pixel reconstruction methods, HQ-JEPA learns semantic representations directly in latent space and uses quantum state-overlap-based similarity as an additional regularization signal. We evaluate the pretrained encoder on GeoBench classification and segmentation tasks under linear probing and fine-tuning settings. Results show that HQ-JEPA achieves competitive and often superior performance over strong self-supervised and remote sensing foundation-model baselines, demonstrating the benefit of integrating predictive self-supervision, cross-modal geometric regularization, and quantum fidelity-based representation learning for remote sensing applications.

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

1 major / 0 minor

Summary. The manuscript introduces HQ-JEPA, a hybrid quantum-classical joint-embedding predictive architecture for cross-modal remote sensing representation learning from paired Sentinel-1 and Sentinel-2 imagery. It extends JEPA-style masked latent prediction with cross-modal token alignment, SIGReg-based Gaussian regularization in the fused latent space, and a differentiable SWAP-test-based Fidelity Quantum Similarity (FQS) loss. The pretrained encoder is evaluated on GeoBench classification and segmentation tasks under linear probing and fine-tuning, with the abstract claiming competitive or superior performance over self-supervised and remote sensing foundation-model baselines, thereby demonstrating the benefit of integrating predictive self-supervision, cross-modal geometric regularization, and quantum fidelity-based representation learning.

Significance. If the performance gains can be shown to arise from the independent contribution of the quantum fidelity term rather than the classical objectives alone, the work would provide a notable example of incorporating quantum state-overlap measures as regularization in self-supervised remote sensing models. This could encourage further exploration of hybrid quantum-classical techniques in geospatial representation learning, provided the empirical isolation of each component is addressed.

major comments (1)
  1. [Abstract] Abstract: The central claim that the results demonstrate the benefit of integrating ... quantum fidelity-based representation learning rests on the premise that the FQS loss supplies a distinct additive regularization signal. However, the manuscript provides no controlled ablation (full model vs. identical architecture with the FQS term removed) on the same GeoBench splits, so any observed superiority could be explained entirely by the JEPA-style prediction, cross-modal alignment, or SIGReg objectives; the quantum component's contribution remains unmeasured.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for the detailed and constructive review. We agree that the current manuscript does not isolate the contribution of the FQS loss through a controlled ablation and that this limits the strength of claims about the quantum component. We will add the requested ablation study in the revision.

read point-by-point responses

  1. Referee: [Abstract] Abstract: The central claim that the results demonstrate the benefit of integrating ... quantum fidelity-based representation learning rests on the premise that the FQS loss supplies a distinct additive regularization signal. However, the manuscript provides no controlled ablation (full model vs. identical architecture with the FQS term removed) on the same GeoBench splits, so any observed superiority could be explained entirely by the JEPA-style prediction, cross-modal alignment, or SIGReg objectives; the quantum component's contribution remains unmeasured.

    Authors: We concur that the absence of a controlled ablation isolating the FQS term prevents a definitive attribution of performance gains to the quantum fidelity loss. In the revised manuscript we will add a direct comparison of the full HQ-JEPA model against an otherwise identical architecture trained without the FQS loss, using the same GeoBench splits, training protocol, and evaluation settings. The results will be reported in a new table and discussed in the experimental section, allowing readers to assess the independent regularization effect of the SWAP-test-based fidelity term. revision: yes

Circularity Check

0 steps flagged

No circularity; architecture and claims are empirically grounded

full rationale

The paper introduces a composite loss with four terms (latent prediction, cross-modal alignment, SIGReg, FQS) and reports empirical results on GeoBench under linear probing and fine-tuning. No equation or claim reduces a derived quantity to a fitted input by construction, no self-citation is load-bearing for a uniqueness result, and no ansatz is smuggled via prior work. The central performance claim rests on external benchmark comparisons rather than tautological redefinition of inputs.

Axiom & Free-Parameter Ledger

1 free parameters · 1 axioms · 1 invented entities

Abstract-only review prevents exhaustive enumeration; the model rests on the unverified premise that a differentiable quantum fidelity term adds value and on standard assumptions of self-supervised representation learning.

free parameters (1)
  • loss weighting coefficients
    Four complementary objectives are combined; relative weights are required but not specified in the abstract.
axioms (1)
  • domain assumption A differentiable approximation to quantum state overlap via SWAP test can serve as a useful similarity regularizer in latent space.
    Invoked to justify the FQS loss term.
invented entities (1)
  • Fidelity Quantum Similarity (FQS) loss no independent evidence
    purpose: Additional regularization signal based on quantum state overlap
    Introduced as a novel component; no independent evidence outside the paper is provided in the abstract.

pith-pipeline@v0.9.1-grok · 5730 in / 1310 out tokens · 21944 ms · 2026-06-28T23:17:14.482949+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

56 extracted references · 17 canonical work pages · 8 internal anchors

  1. [1]

    Self-supervised learning from im- ages with a joint-embedding predictive architecture

    Mahmoud Assran, Quentin Duval, Ishan Misra, Piotr Bojanowski, Pascal Vincent, Michael Rabbat, Yann LeCun, and Nicolas Ballas. Self-supervised learning from im- ages with a joint-embedding predictive architecture. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 15619–15629, 2023

    2023

  2. [2]

    Anysat: One earth observation model for many resolutions, scales, and modalities

    Guillaume Astruc, Nicolas Gonthier, Clement Mallet, and Loic Landrieu. Anysat: One earth observation model for many resolutions, scales, and modalities. InProceedings of the Computer Vision and Pattern Recognition Conference, pages 19530–19540, 2025

    2025

  3. [3]

    LeJEPA: Provable and Scalable Self-Supervised Learning Without the Heuristics

    Randall Balestriero and Yann LeCun. Lejepa: Provable and scalable self-supervised learning without the heuristics.arXiv preprint, arXiv:2511.08544, 10, November 2025

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  4. [4]

    BEiT: BERT Pre-Training of Image Transformers

    Hangbo Bao, Li Dong, Songhao Piao, and Furu Wei. Beit: Bert pre-training of image transformers.arXiv preprint arXiv:2106.08254, 2021

    work page internal anchor Pith review Pith/arXiv arXiv 2021

  5. [5]

    VICReg: Variance-Invariance-Covariance Regularization for Self-Supervised Learning

    Adrien Bardes, Jean Ponce, and Yann LeCun. Vicreg: Variance-invariance-covariance regularization for self-supervised learning.arXiv preprint arXiv:2105.04906, 2021

    work page internal anchor Pith review Pith/arXiv arXiv 2021

  6. [6]

    Satlaspretrain: A large-scale dataset for remote sensing image understanding

    Favyen Bastani, Piper Wolters, Ritwik Gupta, Joe Ferdinando, and Aniruddha Kemb- havi. Satlaspretrain: A large-scale dataset for remote sensing image understanding. In Proceedings of the IEEE/CVF ICCV, pages 16772–16782, 2023

    2023

  7. [7]

    Quantum machine learning.Nature, 549(7671):195–202, 2017

    Jacob Biamonte, Peter Wittek, Nicola Pancotti, Patrick Rebentrost, Nathan Wiebe, and Seth Lloyd. Quantum machine learning.Nature, 549(7671):195–202, 2017

    2017

  8. [8]

    A survey on joint embedding predictive architectures and world models.Available at SSRN 5772122, 2025

    Shamyo Brotee, Gaurab Chhetri, Sazzad Bin Bashar Polock, Venkata Surya Bel- lamkonda, Amir Rafe, and Subasish Das. A survey on joint embedding predictive architectures and world models.Available at SSRN 5772122, 2025

    2025

  9. [9]

    Emerging properties in self-supervised vision trans- formers

    Mathilde Caron, Hugo Touvron, Ishan Misra, Hervé Jégou, Julien Mairal, Piotr Bo- janowski, and Armand Joulin. Emerging properties in self-supervised vision trans- formers. InProceedings of the IEEE/CVF international conference on computer vision, pages 9650–9660, 2021

    2021

  10. [10]

    Variational quantum algorithms.Nature Reviews Physics, 3(9):625–644, 2021

    Marco Cerezo, Andrew Arrasmith, Ryan Babbush, Simon C Benjamin, Suguru Endo, Keisuke Fujii, Jarrod R McClean, Kosuke Mitarai, Xiao Yuan, Lukasz Cincio, et al. Variational quantum algorithms.Nature Reviews Physics, 3(9):625–644, 2021

    2021

  11. [11]

    Why do we need large batchsizes in contrastive learning? a gradient-bias perspective.Advances in Neural Information Processing Systems, 35:33860–33875, 2022

    Changyou Chen, Jianyi Zhang, Yi Xu, Liqun Chen, Jiali Duan, Yiran Chen, Son Tran, Belinda Zeng, and Trishul Chilimbi. Why do we need large batchsizes in contrastive learning? a gradient-bias perspective.Advances in Neural Information Processing Systems, 35:33860–33875, 2022. 16HOSSAIN ET AL.: HQ-JEPA: HYBRID QUANTUM JEPA FOR REMOTE SENSING

    2022

  12. [12]

    Self- supervised learning for few-shot image classification

    Da Chen, Yuefeng Chen, Yuhong Li, Feng Mao, Yuan He, and Hui Xue. Self- supervised learning for few-shot image classification. InIEEE ICASSP, pages 1745–

  13. [13]

    Vl-jepa: Joint embedding pre- dictive architecture for vision-language, 2026

    Delong Chen, Mustafa Shukor, Theo Moutakanni, Willy Chung, Jade Yu, Te- jaswi Kasarla, Yejin Bang, Allen Bolourchi, Yann LeCun, and Pascale Fung. Vl- jepa: Joint embedding predictive architecture for vision-language.arXiv preprint arXiv:2512.10942, 2025

    work page arXiv 2025

  14. [14]

    A simple framework for contrastive learning of visual representations

    Ting Chen, Simon Kornblith, Mohammad Norouzi, and Geoffrey Hinton. A simple framework for contrastive learning of visual representations. InInternational confer- ence on machine learning, pages 1597–1607. PmLR, 2020

    2020

  15. [15]

    Exploring simple siamese representation learning

    Xinlei Chen and Kaiming He. Exploring simple siamese representation learning. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 15750–15758, 2021

    2021

  16. [16]

    Satmae: Pre-training transformers for temporal and multi-spectral satellite imagery.Advances in Neural Information Pro- cessing Systems, 35:197–211, 2022

    Yezhen Cong, Samar Khanna, Chenlin Meng, Patrick Liu, Erik Rozi, Yutong He, Mar- shall Burke, David Lobell, and Stefano Ermon. Satmae: Pre-training transformers for temporal and multi-spectral satellite imagery.Advances in Neural Information Pro- cessing Systems, 35:197–211, 2022

    2022

  17. [17]

    The geometry of self-supervised learning models and its impact on transfer learning.arXiv preprint arXiv:2209.08622, 2022

    Romain Cosentino, Sarath Shekkizhar, Mahdi Soltanolkotabi, Salman Avestimehr, and Antonio Ortega. The geometry of self-supervised learning models and its impact on transfer learning.arXiv preprint arXiv:2209.08622, 2022

    work page arXiv 2022

  18. [18]

    The hidden pitfalls of the cosine similarity loss.arXiv preprint arXiv:2406.16468, 2024

    Andrew Draganov, Sharvaree Vadgama, and Erik J Bekkers. The hidden pitfalls of the cosine similarity loss.arXiv preprint arXiv:2406.16468, 2024

    work page arXiv 2024

  19. [19]

    Whitening for self-supervised representation learning

    Aleksandr Ermolov, Aliaksandr Siarohin, Enver Sangineto, and Nicu Sebe. Whitening for self-supervised representation learning. InInternational conference on machine learning, pages 3015–3024. PMLR, 2021

    2021

  20. [20]

    Chal- lenges and proposed solutions in modeling multimodal data: A systematic review

    Maryam Farhadizadeh, Maria Weymann, Michael Blaß, Johann Kraus, Christopher Gundler, Sebastian Walter, Noah Hempen, Harald Binder, and Nadine Binder. Chal- lenges and proposed solutions in modeling multimodal data: A systematic review. arXiv preprint arXiv:2505.06945, 2025

    work page arXiv 2025

  21. [21]

    Bootstrap your own latent-a new approach to self-supervised learning.Advances in neural information processing systems, 33:21271–21284, 2020

    Jean-Bastien Grill, Florian Strub, Florent Altché, Corentin Tallec, Pierre Richemond, Elena Buchatskaya, Carl Doersch, Bernardo Avila Pires, Zhaohan Guo, Mohammad Gheshlaghi Azar, et al. Bootstrap your own latent-a new approach to self-supervised learning.Advances in neural information processing systems, 33:21271–21284, 2020

    2020

  22. [22]

    A survey on self-supervised learning: Algorithms, applications, and future trends.IEEE T-PAMI, 46(12):9052–9071, 2024

    Jie Gui, Tuo Chen, Jing Zhang, Qiong Cao, Zhenan Sun, Hao Luo, and Dacheng Tao. A survey on self-supervised learning: Algorithms, applications, and future trends.IEEE T-PAMI, 46(12):9052–9071, 2024

    2024

  23. [23]

    Exploring the gap between collapsed & whitened features in self-supervised learning

    Bobby He and Mete Ozay. Exploring the gap between collapsed & whitened features in self-supervised learning. InInternational Conference on Machine Learning, pages 8613–8634. PMLR, 2022. HOSSAIN ET AL.: HQ-JEPA: HYBRID QUANTUM JEPA FOR REMOTE SENSING17

    2022

  24. [24]

    Momentum con- trast for unsupervised visual representation learning

    Kaiming He, Haoqi Fan, Yuxin Wu, Saining Xie, and Ross Girshick. Momentum con- trast for unsupervised visual representation learning. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 9729–9738, 2020

    2020

  25. [25]

    Masked autoencoders are scalable vision learners

    Kaiming He, Xinlei Chen, Saining Xie, Yanghao Li, Piotr Dollár, and Ross Girshick. Masked autoencoders are scalable vision learners. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 16000–16009, 2022

    2022

  26. [26]

    Self-supervised representation learning for bayesian quantum architecture search

    Zhimin He, Hongxiang Chen, Yan Zhou, Haozhen Situ, Yongyao Li, and Lvzhou Li. Self-supervised representation learning for bayesian quantum architecture search. Physical Review A, 111(3):032403, 2025

    2025

  27. [27]

    Masked image modeling: A survey.International Journal of Computer Vision, 133(10):7154–7200, 2025

    Vlad Hondru, Florinel Alin Croitoru, Shervin Minaee, Radu Tudor Ionescu, and Nicu Sebe. Masked image modeling: A survey.International Journal of Computer Vision, 133(10):7154–7200, 2025

    2025

  28. [28]

    QMC-Net: Data-Aware Quantum Representations for Remote Sensing Image Classification

    Md Aminur Hossain, Ayush V Patel, and Biplab Banerjee. Qmc-net: Data-aware quantum representations for remote sensing image classification.arXiv preprint arXiv:2604.11817, 2026

    work page internal anchor Pith review Pith/arXiv arXiv 2026

  29. [29]

    HQF-Net: A Hybrid Quantum-Classical Multi-Scale Fusion Network for Remote Sensing Image Segmentation

    Md Aminur Hossain, Ayush V Patel, Siddhant Gole, Sanjay K Singh, and Biplab Baner- jee. Hqf-net: A hybrid quantum-classical multi-scale fusion network for remote sensing image segmentation.arXiv preprint arXiv:2604.06715, 2026

    work page internal anchor Pith review Pith/arXiv arXiv 2026

  30. [30]

    HQ-UNet: A Hybrid Quantum-Classical U-Net with a Quantum Bottleneck for Remote Sensing Image Segmentation

    Md Aminur Hossain, Ayush V Patel, Ikshwaku Vanani, and Biplab Banerjee. Hq-unet: A hybrid quantum-classical u-net with a quantum bottleneck for remote sensing image segmentation.arXiv preprint arXiv:2604.27206, 2026

    work page internal anchor Pith review Pith/arXiv arXiv 2026

  31. [31]

    A survey of self-supervised and few-shot object detection.IEEE T-PAMI, 45 (4):4071–4089, 2022

    Gabriel Huang, Issam Laradji, David Vazquez, Simon Lacoste-Julien, and Pau Ro- driguez. A survey of self-supervised and few-shot object detection.IEEE T-PAMI, 45 (4):4071–4089, 2022

    2022

  32. [32]

    Self-supervised learning for medical image classification: a systematic review and implementation guidelines.NPJ Digital Medicine, 6(1):74, 2023

    Shih-Cheng Huang, Anuj Pareek, Malte Jensen, Matthew P Lungren, Serena Yeung, and Akshay S Chaudhari. Self-supervised learning for medical image classification: a systematic review and implementation guidelines.NPJ Digital Medicine, 6(1):74, 2023

    2023

  33. [33]

    Quantum self-supervised learning.Quantum Science & Technology, 7 (3):035005, 2022

    Ben Jaderberg, Lewis W Anderson, Weidi Xie, Samuel Albanie, Martin Kiffner, and Dieter Jaksch. Quantum self-supervised learning.Quantum Science & Technology, 7 (3):035005, 2022

    2022

  34. [34]

    A survey on contrastive self-supervised learning.Technologies, 9 (1):2, 2020

    Ashish Jaiswal, Ashwin Ramesh Babu, Mohammad Zaki Zadeh, Debapriya Banerjee, and Fillia Makedon. A survey on contrastive self-supervised learning.Technologies, 9 (1):2, 2020

    2020

  35. [35]

    Understanding dimensional collapse in contrastive self-supervised learning.arXiv preprint arXiv:2110.09348, 2021

    Li Jing, Pascal Vincent, Yann LeCun, and Yuandong Tian. Understanding dimensional collapse in contrastive self-supervised learning.arXiv preprint arXiv:2110.09348, 2021

    work page arXiv 2021

  36. [36]

    Debanjan Konar, Siddhartha Bhattacharyya, Bijaya K Panigrahi, and Elizabeth C Behrman. Qutrit-inspired fully self-supervised shallow quantum learning network for brain tumor segmentation.IEEE Transactions on Neural Networks and Learning Sys- tems, 33(11):6331–6345, 2021. 18HOSSAIN ET AL.: HQ-JEPA: HYBRID QUANTUM JEPA FOR REMOTE SENSING

    2021

  37. [37]

    Geo-bench: Toward foundation models for earth monitoring.Advances in Neural Information Processing Systems, 36:51080–51093, 2023

    Alexandre Lacoste, Nils Lehmann, Pau Rodriguez, Evan Sherwin, Hannah Kerner, Björn Lütjens, Jeremy Irvin, David Dao, Hamed Alemohammad, Alexandre Drouin, et al. Geo-bench: Toward foundation models for earth monitoring.Advances in Neural Information Processing Systems, 36:51080–51093, 2023

    2023

  38. [38]

    Qsea: Quantum self- supervised learning with entanglement augmentation.Advanced Quantum Technolo- gies, 9(4):e00530, 2026

    LingXiao Li, XiaoHui Ni, Jing Li, SuJuan Qin, and Fei Gao. Qsea: Quantum self- supervised learning with entanglement augmentation.Advanced Quantum Technolo- gies, 9(4):e00530, 2026

    2026

  39. [39]

    Quantum fidelity measures for mixed states.Reports on Progress in Physics, 82(7):076001, 2019

    Yeong-Cherng Liang, Yu-Hao Yeh, Paulo EMF Mendonça, Run Yan Teh, Margaret D Reid, and Peter D Drummond. Quantum fidelity measures for mixed states.Reports on Progress in Physics, 82(7):076001, 2019

    2019

  40. [40]

    V-JEPA 2.1: Unlocking Dense Features in Video Self-Supervised Learning

    Lorenzo Mur-Labadia, Matthew Muckley, Amir Bar, Mido Assran, Koustuv Sinha, Mike Rabbat, Yann LeCun, Nicolas Ballas, and Adrien Bardes. V-jepa 2.1: Unlocking dense features in video self-supervised learning.arXiv preprint arXiv:2603.14482, 2026

    work page internal anchor Pith review Pith/arXiv arXiv 2026

  41. [41]

    Simreg: Regression as a simple yet effective tool for self-supervised knowledge distillation.arXiv preprint arXiv:2201.05131, 2022

    KL Navaneet, Soroush Abbasi Koohpayegani, Ajinkya Tejankar, and Hamed Pirsi- avash. Simreg: Regression as a simple yet effective tool for self-supervised knowledge distillation.arXiv preprint arXiv:2201.05131, 2022

    work page arXiv 2022

  42. [42]

    Mmearth: Exploring multi-modal pretext tasks for geospatial repre- sentation learning

    Vishal Nedungadi, Ankit Kariryaa, Stefan Oehmcke, Serge Belongie, Christian Igel, and Nico Lang. Mmearth: Exploring multi-modal pretext tasks for geospatial repre- sentation learning. InECCV, pages 164–182. Springer, 2024

    2024

  43. [43]

    Self-supervised learning for few-shot medical image segmentation.IEEE Transactions on Medical Imaging, 41(7):1837–1848, 2022

    Cheng Ouyang, Carlo Biffi, Chen Chen, Turkay Kart, Huaqi Qiu, and Daniel Rueckert. Self-supervised learning for few-shot medical image segmentation.IEEE Transactions on Medical Imaging, 41(7):1837–1848, 2022

    2022

  44. [44]

    Qclr: Quantum-lstm contrastive learning frame- work for continuous mental health monitoring.Expert Systems with Applications, 238: 121921, 2024

    Anupama Padha and Anita Sahoo. Qclr: Quantum-lstm contrastive learning frame- work for continuous mental health monitoring.Expert Systems with Applications, 238: 121921, 2024

    2024

  45. [45]

    Scale- mae: A scale-aware masked autoencoder for multiscale geospatial representation learn- ing

    Colorado J Reed, Ritwik Gupta, Shufan Li, Sarah Brockman, Christopher Funk, Brian Clipp, Kurt Keutzer, Salvatore Candido, Matt Uyttendaele, and Trevor Darrell. Scale- mae: A scale-aware masked autoencoder for multiscale geospatial representation learn- ing. InProceedings of the IEEE/CVF ICCV, pages 4088–4099, 2023

    2023

  46. [46]

    Quantum con- trastive learning for human activity recognition.Smart Health, 36:100574, 2025

    Yanhui Ren, Di Wang, Lingling An, Shiwen Mao, and Xuyu Wang. Quantum con- trastive learning for human activity recognition.Smart Health, 36:100574, 2025

    2025

  47. [47]

    Collapse-proof non- contrastive self-supervised learning.arXiv preprint arXiv:2410.04959, 2024

    Emanuele Sansone, Tim Lebailly, and Tinne Tuytelaars. Collapse-proof non- contrastive self-supervised learning.arXiv preprint arXiv:2410.04959, 2024

    work page arXiv 2024

  48. [48]

    DINOv3

    Oriane Siméoni, Huy V V o, Maximilian Seitzer, Federico Baldassarre, Maxime Oquab, Cijo Jose, Vasil Khalidov, Marc Szafraniec, Seungeun Yi, Michaël Ramamonjisoa, et al. Dinov3.arXiv preprint arXiv:2508.10104, 2025

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  49. [49]

    Prithvi-eo-2.0: A versatile multi-temporal foun- dation model for earth observation applications.IEEE TGRS, 2025

    Daniela Szwarcman, Sujit Roy, Paolo Fraccaro, Orsteinn Elí Gíslason, Benedikt Blu- menstiel, Rinki Ghosal, Pedro Henrique De Oliveira, Joao Lucas de Sousa Almeida, Rocco Sedona, Yanghui Kang, et al. Prithvi-eo-2.0: A versatile multi-temporal foun- dation model for earth observation applications.IEEE TGRS, 2025. HOSSAIN ET AL.: HQ-JEPA: HYBRID QUANTUM JEPA...

    2025

  50. [50]

    Decoupling common and unique representations for multimodal self-supervised learning

    Yi Wang, Conrad M Albrecht, Nassim Ait Ali Braham, Chenying Liu, Zhitong Xiong, and Xiao Xiang Zhu. Decoupling common and unique representations for multimodal self-supervised learning. InECCV, pages 286–303. Springer, 2024

    2024

  51. [51]

    Neural plasticity-inspired foundation model for observing the Earth crossing modalities.arXiv preprint arXiv:2403.15356, 2024

    Zhitong Xiong, Yi Wang, Fahong Zhang, Adam J Stewart, Joëlle Hanna, Damian Borth, Ioannis Papoutsis, Bertrand Le Saux, Gustau Camps-Valls, and Xiao Xiang Zhu. Neural plasticity-inspired multimodal foundation model for earth observation. arXiv preprint arXiv:2403.15356, 2024

    work page arXiv 2024

  52. [52]

    Self-supervised pre-trained neural network for quantum natural language processing.Neural Networks, 184:107004, 2025

    Ben Yao, Prayag Tiwari, and Qiuchi Li. Self-supervised pre-trained neural network for quantum natural language processing.Neural Networks, 184:107004, 2025

    2025

  53. [53]

    Contrastive learning with large memory bank and negative embedding subtraction for accurate copy detection.arXiv preprint arXiv:2112.04323, 2021

    Shuhei Yokoo. Contrastive learning with large memory bank and negative embedding subtraction for accurate copy detection.arXiv preprint arXiv:2112.04323, 2021

    work page arXiv 2021

  54. [54]

    QUSL: Quantum Unsupervised Image Similarity Learning with Enhanced Perfor- mance.arXiv e-prints, April 2024

    Lian-Hui Yu, Xiao-Yu Li, Geng Chen, Qin-Sheng Zhu, Hui Li, and Guo-Wu Yang. QUSL: Quantum Unsupervised Image Similarity Learning with Enhanced Perfor- mance.arXiv e-prints, April 2024

    2024

  55. [55]

    Barlow twins: Self-supervised learning via redundancy reduction

    Jure Zbontar, Li Jing, Ishan Misra, Yann LeCun, and Stéphane Deny. Barlow twins: Self-supervised learning via redundancy reduction. InInternational conference on ma- chine learning, pages 12310–12320. PMLR, 2021

    2021

  56. [56]

    Self-supervised mul- timodal learning: A survey.IEEE Transactions on Pattern Analysis and Machine In- telligence, 47(7):5299–5318, 2024

    Yongshuo Zong, Oisin Mac Aodha, and Timothy M Hospedales. Self-supervised mul- timodal learning: A survey.IEEE Transactions on Pattern Analysis and Machine In- telligence, 47(7):5299–5318, 2024

    2024