Pith. sign in

REVIEW 3 major objections 2 minor 9 cited by

Quantum-Enhanced Optimization by Warm Starts

T0 review · 3 major / 2 minor · reviewed 2026-08-05 · deepseek-v4-flash

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

Pith's one-line read This paper claims that quantum-generated samples, used as warm starts, can reduce the total runtime of classical heuristics on Max-Cut and Maximum Independent Set, even on real quantum hardware.

desk verdict The supplied full text is a different paper, so the quantum warm-start runtime claim can only be judged at the level of the abstract—which is plausible but unverifiable. read the letter →

arxiv 2508.16309 v1 pith:74BV3TSX submitted 2025-08-22 quant-ph

Ieva Čepaitė , Niam Vaishnav , Leo Zhou , Ashley Montanaro This is my paper
classification quant-ph
keywords quantum-enhancedoptimizationQAOAwarmstartsMax-CutMaximumIndependentSetcombinatorialerrormitigationqubitmapping
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

The paper tries to show that a hybrid pipeline—run a quantum approximate optimization algorithm (QAOA) briefly, then feed its output samples as initial guesses to a classical solver—can beat the classical solver starting from scratch. The target problems are hard combinatorial optimization tasks such as Max-Cut and Maximum Independent Set. The authors claim experimental evidence, including on quantum hardware, that the combined wall-clock time to a good solution is lower than the original classical algorithm's. If this holds, quantum sampling would serve as a practical accelerator for existing classical heuristics, not as a standalone solver.

What carries the argument

The warm start itself is the key object: a distribution of candidate solutions sampled from QAOA, a hybrid quantum-classical algorithm that encodes the optimization objective as a Hamiltonian and alternates problem-dependent and mixing unitaries. The quantum samples replace random or heuristic initialization, giving the classical heuristic a biased starting set that is closer to good solutions. Supporting machinery includes low-cost QAOA parameter heuristics, qubit mapping and routing to reduce circuit depth, and error mitigation to preserve sample quality.

What would settle it

Run both the quantum-enhanced pipeline and the original classical heuristic on a fixed set of Max-Cut and MIS instances, measuring total wall-clock time to reach a target approximation quality, with quantum overhead fully counted; if the enhanced pipeline does not achieve a lower median time on any instance class, the central claim is refuted. A complementary control: replace the quantum samples with uniformly random samples of the same size drawn at the same cost; equal performance would show the warm-start signal is not quantum-specific.

Watch

Extended reading notes

Core claim

The central claim is that quantum-generated samples can serve as warm starts that put classical heuristics in a better initial basin, so they converge to high-quality solutions faster than from default starts. The paper introduces parameter-setting strategies for QAOA that avoid expensive optimization loops, along with qubit mapping, routing, and error-mitigation techniques that reduce gate counts and noise. Experiments reported in the abstract show runtime improvements for Max-Cut and MIS, including on quantum hardware, against the original classical algorithms.

Load-bearing premise

The full pipeline wins only if the total cost of producing the quantum warm-start samples—including QAOA parameter selection, qubit mapping, routing, error mitigation, and hardware noise—is smaller than the time the classical heuristic saves by starting from those samples.

Editorial extensions

If this is right

  • Classical optimization heuristics for Max-Cut and MIS can be accelerated without redesigning them, by prepending a short quantum sampling stage.
  • The engineering techniques that reduce QAOA overhead (parameter setting, routing, error mitigation) are load-bearing: without them, the warm-start gain is likely erased by sampling cost.
  • The approach targets near-term hardware, suggesting near-term quantum devices can have a practical role in combinatorial optimization before fault tolerance.
  • If runtime gains hold on real hardware, the method provides a benchmark for when quantum sampling is actually useful for classical workflows.

Reading between the lines

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

  • The same warm-start mechanism should transfer to other quantum samplers and other classical heuristics, provided the sample distribution is sufficiently biased toward good solutions.
  • The win is likely instance-dependent: QAOA's bias is stronger on instances with structure correlated with the problem Hamiltonian, so a fair comparison should report performance across instance classes.
  • A direct test of whether QAOA is the cause: replace QAOA samples with uniform random samples of equal size at the same cost; if the runtime gain disappears, the quantum bias is the active ingredient.
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

3 major / 2 minor

Summary. The submission as received consists of the abstract for arXiv:2508.16309, which proposes a "quantum-enhanced optimization" pipeline: QAOA-generated samples are used as warm starts for classical Max-Cut and Maximum Independent Set heuristics, supported by novel QAOA parameter-setting strategies, qubit mapping/routing techniques, and error mitigation, with claimed runtime improvements on quantum hardware. However, the provided full text is arXiv:2508.16317 (cs.CV), a position paper on image-size-agnostic, task-driven vision encoders. The body contains no equations, algorithms, experiments, or technical content related to the abstract. The only reviewable material is the abstract, which asserts the central runtime improvement claim without quantitative support.

Significance. If the claimed speedups are substantiated, warm-starting classical combinatorial heuristics with QAOA samples at acceptable total wall-clock cost would be a practically meaningful result for quantum-enhanced optimization, especially with hardware demonstrations. However, because the body of the submission is a different paper, the technical content necessary to assess this contribution is entirely absent. The central claim is therefore unverifiable in the present form, and the paper cannot be evaluated on its merits.

major comments (3)
  1. [Full Text] The provided full text is arXiv:2508.16317, a computer vision position paper on image-size-agnostic encoders. It contains no discussion of QAOA, Max-Cut, MIS, quantum hardware, or runtime comparisons. The central claim of the abstract is thus entirely unsupported by the body. This is not a local presentation issue; the manuscript as submitted is not reviewable for its stated topic.
  2. [Abstract] The runtime claim "Experimental results ... showcase runtime improvements" is asserted with no methodological detail. It is unspecified whether the comparison includes the full quantum pipeline overhead: QAOA variational parameter optimization, qubit mapping/routing, compilation, error mitigation, and hardware access time. It is also unspecified what "original classical algorithms" means precisely (e.g., same heuristic with default initialization and same termination criterion). Without these definitions and a timing breakdown, no speedup claim can be audited.
  3. [Abstract] The "novel parameter-setting strategies" for QAOA are not described anywhere in the provided manuscript. This matters because the generic risk in this line of work is that parameter-setting rules are tuned on the same benchmark instances used to report speedups, which would make the improvement a fitted outcome rather than a generalizable property. The manuscript provides no way to check this.
minor comments (2)
  1. [Full Text] The reference list and acknowledgments correspond to the vision paper (arXiv:2508.16317) and are irrelevant to the abstract of arXiv:2508.16309. This further indicates that the supplied body does not belong with the abstract.
  2. [Abstract] The phrase "quantum-enhanced optimization" would benefit from explicit comparison with existing quantum-warm-start and hybrid quantum-classical literature, but this is secondary to the missing body.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity identifiable: the abstract alone provides no derivation chain, and the supplied full text is a different paper (arXiv:2508.16317).

full rationale

The only text belonging to the claimed paper (arXiv:2508.16309) is its abstract. The attached full text is arXiv:2508.16317, a computer-vision position paper by different authors, so none of the quantum warm-start claims—QAOA parameter-setting strategies, qubit mapping/routing, error mitigation, or runtime improvements—can be traced through equations, algorithms, or experimental baselines. Circularity analysis requires exhibiting a specific reduction such as an equation that is equivalent to its input by construction, or a fitted parameter renamed as a prediction. No such reduction is visible in the abstract. The abstract's runtime claim is unverifiable from the provided materials, and the full-text mismatch is a completeness/integrity concern rather than evidence of circularity. Unverifiability is not circularity. Therefore, the honest finding is no significant circularity identified, score 0.

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

The abstract-level review rests entirely on the empirical claim that quantum sampling plus classical refinement beats the classical baseline. Free parameters: QAOA angles, set by an unpublished parameter-setting strategy, are the main tunable knobs whose values are not disclosed, and the classical heuristic stopping budget is unstated. Domain assumptions: hardware sampling must be fast and reliable enough to beat the classical-only path, and QAOA samples must be informative as warm starts. No invented entities.

free parameters (2)
  • QAOA variational angles produced by the paper's parameter-setting strategy = not given in abstract
    The abstract advertises novel parameter-setting strategies for QAOA; the angles are the degrees of freedom tuned to make sampling useful. Their values are not disclosed, and an instance-specific tuning loop would count as fitted parameters.
  • Classical heuristic runtime budget or stopping criterion = not given in abstract
    The claimed runtime comparison requires a defined stopping rule for the classical heuristics; the abstract does not state it, and the reported speedup ratio depends sensitively on it.
assumptions (2)
  • domain assumption Near-term quantum hardware can produce useful QAOA samples quickly enough and with enough signal to warm-start a classical heuristic within the total runtime budget.
    The entire runtime-win claim depends on sampling plus overhead beating the classical-only path. Invoked implicitly by 'Experimental results... showcase runtime improvements'.
  • domain assumption Quantum-generated warm starts carry more structure than random or trivial initial solutions for the classical Max-Cut and MIS heuristics used.
    The premise that warm starts accelerate the classical algorithms requires the quantum samples to be informative initial solutions for these particular heuristics.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Quantum-Enhanced Optimization by Warm Starts." pith.science (2026). https://pith.science/paper/74BV3TSX

@misc{pith2026250816309,
  author       = {Pith},
  title        = {Pith review of: Quantum-Enhanced Optimization by Warm Starts},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/74BV3TSX}},
  note         = {Machine review of arXiv:2508.16309}
}
read the original abstract

We present an approach, which we term quantum-enhanced optimization, to accelerate classical optimization algorithms by leveraging quantum sampling. Our method uses quantum-generated samples as warm starts to classical heuristics for solving challenging combinatorial problems like Max-Cut and Maximum Independent Set (MIS). To implement the method efficiently, we introduce novel parameter-setting strategies for the Quantum Approximate Optimization Algorithm (QAOA), qubit mapping and routing techniques to reduce gate counts, and error-mitigation techniques. Experimental results, including on quantum hardware, showcase runtime improvements compared with the original classical algorithms.

Discussion (0). Sign in to comment.

Forward citations

Cited by 9 Pith papers

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

  1. Constraint-Aware Quantum Optimization via Hamming Weight Operators

    quant-ph 2026-01 unverdicted novelty 8.0 of 10

    Hamming Weight Operators and an adaptive QAOA variant confine evolution to feasible states by construction, delivering faster convergence and roughly half the gate count versus penalty methods on finance and physics tasks.

  2. Quantum-informed surrogate sampling for combinatorial optimization

    quant-ph 2026-07 conditional novelty 6.0 of 10

    QISS classically samples a pairwise model built from O(N) low-weight QAOA correlators and outperforms standard QAOA at larger depths on MaxCut and MIS benchmarks.

  3. Quantum Approximate Optimization via Noise-Directed Adaptive Warm-Starting

    quant-ph 2026-07 conditional novelty 6.0 of 10

    Bitflip-gauge warm-start QAOA that aligns the ansatz with amplitude-damping noise improves 100-qubit Ising approximation ratios over non-gauge iterative warm-start at no extra circuit cost.

  4. Quantum-Informed Portfolio Selection: An End-to-End Pipeline Validated on Trapped-Ion Hardware with Real Market Data

    quant-ph 2026-07 conditional novelty 6.0 of 10

    qReduMIS hybrid pipeline improves QAOA performance on real financial MIS instances up to 225 assets, achieving higher success probabilities and better scaling on Quantinuum trapped-ion hardware.

  5. Quantum Elastic Network Models and their Application to Graphene

    quant-ph 2026-01 conditional novelty 6.0 of 10

    A quantum algorithm for coupled oscillators is adapted to elastic network models, with an efficient connectivity oracle for graphene and applications to heat transfer and rippling — at the cost of a coarse two-bucket ...

  6. Quantum-Informed Portfolio Selection: An End-to-End Pipeline Validated on Trapped-Ion Hardware with Real Market Data

    quant-ph 2026-07 conditional novelty 5.5 of 10

    qReduMIS, using QAOA frozen-node signals plus classical reductions, solves real market MIS portfolio instances up to 225 assets on Helios with far better success and TTS scaling than standalone QAOA.

  7. Mind the gaps: The fraught road to quantum advantage

    quant-ph 2025-10 unverdicted novelty 4.0 of 10

    The authors identify four transitions needed to reach fault-tolerant application-scale quantum computing from current NISQ devices.

  8. Setting angles in quantum approximate optimization at utility-scale

    quant-ph 2026-06 unverdicted novelty 3.0 of 10

    The paper benchmarks approximation techniques and transfer learning for setting QAOA angles at utility scale and extracts operational guidance from hardware-validated results.

  9. Mind the gaps: The fraught road to quantum advantage

    quant-ph 2025-10 unverdicted novelty 3.0 of 10

    The paper identifies four key hurdles in the transition from NISQ to FASQ quantum computers and argues that targeting them will accelerate progress toward useful quantum advantage.

Reference graph

Works this paper leans on

49 extracted references · 16 canonical work pages · cited by 7 Pith papers

  1. [1]

    Multiple object recognition with visual attention

    Jimmy Ba, V olodymyr Mnih, and Koray Kavukcuoglu. Multiple object recognition with visual attention. arXiv preprint arXiv:1412.7755, 2014

    arXiv 2014

  2. [2]

    Recurrent memory transformer.Advances in Neural Information Processing Systems, 35:11079–11091, 2022

    Aydar Bulatov, Yury Kuratov, and Mikhail Burtsev. Recurrent memory transformer.Advances in Neural Information Processing Systems, 35:11079–11091, 2022

    2022

  3. [3]

    Unsupervised Foveal Vision Neural Networks with Top-Down Attention

    Ryan Burt, Nina N Thigpen, Andreas Keil, and Jose C Principe. Unsupervised foveal vision neural networks with top-down attention. arXiv preprint arXiv:2010.09103, 2020

    work page Pith review arXiv 2010

  4. [4]

    End-to-end object detection with transformers

    Nicolas Carion, Francisco Massa, Gabriel Synnaeve, Nicolas Usunier, Alexander Kirillov, and Sergey Zagoruyko. End-to-end object detection with transformers. In European conference on computer vision, pages 213–229. Springer, 2020

    2020

  5. [5]

    Emerging properties in self-supervised vision transformers

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

    2021

  6. [6]

    Crossvit: Cross-attention multi- scale vision transformer for image classification

    Chun-Fu Richard Chen, Quanfu Fan, and Rameswar Panda. Crossvit: Cross-attention multi- scale vision transformer for image classification. In Proceedings of the IEEE/CVF international conference on computer vision, pages 357–366, 2021

    work page 2021

  7. [7]

    Twins: Revisiting the design of spatial attention in vision transformers

    Xiangxiang Chu, Zhi Tian, Yuqing Wang, Bo Zhang, Haibing Ren, Xiaolin Wei, Huaxia Xia, and Chunhua Shen. Twins: Revisiting the design of spatial attention in vision transformers. Advances in neural information processing systems, 34:9355–9366, 2021

    2021

  8. [8]

    Transformer-xl: Attentive language models beyond a fixed-length context

    Zihang Dai, Zhilin Yang, Yiming Yang, Jaime Carbonell, Quoc V Le, and Ruslan Salakhutdinov. Transformer-xl: Attentive language models beyond a fixed-length context. arXiv preprint arXiv:1901.02860, 2019

    arXiv 1901

Show all 49 references
  1. [9]

    Vision transformers need registers

    Timothée Darcet, Maxime Oquab, Julien Mairal, and Piotr Bojanowski. Vision transformers need registers. arXiv preprint arXiv:2309.16588, 2023

    2023 arXiv

  2. [10]

    Imagenet: A large- scale hierarchical image database

    Jia Deng, Wei Dong, Richard Socher, Li-Jia Li, Kai Li, and Li Fei-Fei. Imagenet: A large- scale hierarchical image database. In 2009 IEEE conference on computer vision and pattern recognition, pages 248–255. Ieee, 2009

    2009

  3. [11]

    Bert: Pre-training of deep bidirectional transformers for language understanding

    Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. Bert: Pre-training of deep bidirectional transformers for language understanding. In Proceedings of the 2019 conference of the North American chapter of the association for computational linguistics: human langu...

    2019

  4. [12]

    An image is worth 16x16 words: Transformers for image recognition at scale

    Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xiaohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias Minderer, Georg Heigold, Sylvain Gelly, Jakob Uszkoreit, and Neil Houlsby. An image is worth 16x16 words: Transformers for image recognition at...

    2021

  5. [13]

    Multiscale vision transformers

    Haoqi Fan, Bo Xiong, Karttikeya Mangalam, Yanghao Li, Zhicheng Yan, Jitendra Malik, and Christoph Feichtenhofer. Multiscale vision transformers. In Proceedings of the IEEE/CVF international conference on computer vision, pages 6824–6835, 2021

    2021

  6. [14]

    Levit: a vision transformer in convnet’s clothing for faster inference

    Benjamin Graham, Alaaeldin El-Nouby, Hugo Touvron, Pierre Stock, Armand Joulin, Hervé Jégou, and Matthijs Douze. Levit: a vision transformer in convnet’s clothing for faster inference. In Proceedings of the IEEE/CVF international conference on computer vision, pages 12259– 12269, 2021

    2021

  7. [15]

    Gmat: Global memory augmentation for transformers

    Ankit Gupta and Jonathan Berant. Gmat: Global memory augmentation for transformers. arXiv preprint arXiv:2006.03274, 2020

    2006 arXiv

  8. [16]

    Deep residual learning for image recognition

    Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. Deep residual learning for image recognition. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 770–778, 2016. 10

    2016

  9. [17]

    Rethinking spatial dimensions of vision transformers

    Byeongho Heo, Sangdoo Yun, Dongyoon Han, Sanghyuk Chun, Junsuk Choe, and Seong Joon Oh. Rethinking spatial dimensions of vision transformers. In Proceedings of the IEEE/CVF international conference on computer vision, pages 11936–11945, 2021

    2021

  10. [18]

    Long short-term memory

    Sepp Hochreiter and Jürgen Schmidhuber. Long short-term memory. Neural computation, 9(8):1735–1780, 1997

    1997

  11. [19]

    Foveater: Foveated transformer for image classification

    Aditya Jonnalagadda, William Yang Wang, BS Manjunath, and Miguel P Eckstein. Foveater: Foveated transformer for image classification. arXiv preprint arXiv:2105.14173, 2021

    2021 arXiv

  12. [20]

    Learning multiple layers of features from tiny images

    Alex Krizhevsky, Geoffrey Hinton, et al. Learning multiple layers of features from tiny images. 2009

    2009

  13. [21]

    Imagenet classification with deep convolutional neural networks

    Alex Krizhevsky, Ilya Sutskever, and Geoffrey E Hinton. Imagenet classification with deep convolutional neural networks. Advances in neural information processing systems, 25, 2012

    2012

  14. [22]

    The shape of ai to come! Talk presented at the AI Action Summit 2025, February

    Yann LeCun. The shape of ai to come! Talk presented at the AI Action Summit 2025, February

    2025

  15. [23]

    Gradient-based learning applied to document recognition

    Yann LeCun, Léon Bottou, Yoshua Bengio, and Patrick Haffner. Gradient-based learning applied to document recognition. Proceedings of the IEEE, 86(11):2278–2324, 1998

    1998

  16. [24]

    Swin transformer v2: Scaling up capacity and resolution

    Ze Liu, Han Hu, Yutong Lin, Zhuliang Yao, Zhenda Xie, Yixuan Wei, Jia Ning, Yue Cao, Zheng Zhang, Li Dong, et al. Swin transformer v2: Scaling up capacity and resolution. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 12009–12019, 2022

    2022

  17. [25]

    Swin transformer: Hierarchical vision transformer using shifted windows

    Ze Liu, Yutong Lin, Yue Cao, Han Hu, Yixuan Wei, Zheng Zhang, Stephen Lin, and Baining Guo. Swin transformer: Hierarchical vision transformer using shifted windows. In Proceedings of the IEEE/CVF international conference on computer vision, pages 10012–10022, 2021

    2021

  18. [26]

    Decoupled weight decay regularization

    Ilya Loshchilov and Frank Hutter. Decoupled weight decay regularization. In International Conference on Learning Representations, 2019

    2019

  19. [27]

    Biologically inspired deep learning model for efficient foveal-peripheral vision

    Hristofor Lukanov, Peter König, and Gordon Pipa. Biologically inspired deep learning model for efficient foveal-peripheral vision. Frontiers in Computational Neuroscience, 15:746204, 2021

    2021

  20. [28]

    Implicit-zoo: A large-scale dataset of neural implicit functions for 2d images and 3d scenes, 2024

    Qi Ma, Danda Pani Paudel, Ender Konukoglu, and Luc Van Gool. Implicit-zoo: A large-scale dataset of neural implicit functions for 2d images and 3d scenes, 2024

    2024

  21. [29]

    Recurrent models of visual attention

    V olodymyr Mnih, Nicolas Heess, Alex Graves, and Koray Kavukcuoglu. Recurrent models of visual attention. Advances in neural information processing systems, 27, 2014

    2014

  22. [30]

    A focused backpropagation algorithm for temporal pattern recognition

    Michael C Mozer. A focused backpropagation algorithm for temporal pattern recognition. In Backpropagation, pages 137–169. Psychology Press, 2013

    2013

  23. [31]

    Dinov2: Learning robust visual features without supervision

    Maxime Oquab, Timothée Darcet, Théo Moutakanni, Huy V o, Marc Szafraniec, Vasil Khalidov, Pierre Fernandez, Daniel Haziza, Francisco Massa, Alaaeldin El-Nouby, et al. Dinov2: Learning robust visual features without supervision. arXiv preprint arXiv:2304.07193, 2023

    2023 arXiv

  24. [32]

    Compressive transformers for long-range sequence modelling

    Jack W Rae, Anna Potapenko, Siddhant M Jayakumar, and Timothy P Lillicrap. Compressive transformers for long-range sequence modelling. arXiv preprint arXiv:1911.05507, 2019

    1911 arXiv

  25. [33]

    The utility driven dynamic error propagation network, volume 11

    Anthony J Robinson and Frank Fallside. The utility driven dynamic error propagation network, volume 11. University of Cambridge Department of Engineering Cambridge, 1987

    1987

  26. [34]

    Learning to generate artificial fovea trajectories for target detection

    Juergen Schmidhuber and Rudolf Huber. Learning to generate artificial fovea trajectories for target detection. International Journal of Neural Systems, 2(01n02):125–134, 1991

    1991

  27. [35]

    Proximal policy optimization algorithms

    John Schulman, Filip Wolski, Prafulla Dhariwal, Alec Radford, and Oleg Klimov. Proximal policy optimization algorithms. arXiv preprint arXiv:1707.06347, 2017. 11

    2017 arXiv

  28. [36]

    Deepseekmath: Pushing the limits of mathematical reasoning in open language models

    Zhihong Shao, Peiyi Wang, Qihao Zhu, Runxin Xu, Junxiao Song, Xiao Bi, Haowei Zhang, Mingchuan Zhang, YK Li, Y Wu, et al. Deepseekmath: Pushing the limits of mathematical reasoning in open language models. arXiv preprint arXiv:2402.03300, 2024

    2024 arXiv

  29. [37]

    Very deep convolutional networks for large-scale image recognition

    Karen Simonyan and Andrew Zisserman. Very deep convolutional networks for large-scale image recognition. arXiv preprint arXiv:1409.1556, 2014

    2014 arXiv

  30. [38]

    Going deeper with convolutions

    Christian Szegedy, Wei Liu, Yangqing Jia, Pierre Sermanet, Scott Reed, Dragomir Anguelov, Dumitru Erhan, Vincent Vanhoucke, and Andrew Rabinovich. Going deeper with convolutions. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 1–9, 2015

    2015

  31. [39]

    Training data-efficient image transformers & distillation through attention

    Hugo Touvron, Matthieu Cord, Matthijs Douze, Francisco Massa, Alexandre Sablayrolles, and Hervé Jégou. Training data-efficient image transformers & distillation through attention. In International conference on machine learning, pages 10347–10357. PMLR, 2021

    2021

  32. [40]

    Fixing the train-test resolution discrepancy

    Hugo Touvron, Andrea Vedaldi, Matthijs Douze, and Hervé Jégou. Fixing the train-test resolution discrepancy. Advances in neural information processing systems, 32, 2019

    2019

  33. [41]

    Attention is all you need

    Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N Gomez, Łukasz Kaiser, and Illia Polosukhin. Attention is all you need. Advances in neural information processing systems, 30, 2017

    2017

  34. [42]

    Pyramid vision transformer: A versatile backbone for dense prediction without convolutions

    Wenhai Wang, Enze Xie, Xiang Li, Deng-Ping Fan, Kaitao Song, Ding Liang, Tong Lu, Ping Luo, and Ling Shao. Pyramid vision transformer: A versatile backbone for dense prediction without convolutions. In Proceedings of the IEEE/CVF international conference on computer vision, pa...

    2021

  35. [43]

    Generalization of backpropagation with application to a recurrent gas market model

    Paul J Werbos. Generalization of backpropagation with application to a recurrent gas market model. Neural networks, 1(4):339–356, 1988

    1988

  36. [44]

    Simple statistical gradient-following algorithms for connectionist reinforce- ment learning

    Ronald J Williams. Simple statistical gradient-following algorithms for connectionist reinforce- ment learning. Machine learning, 8:229–256, 1992

    1992

  37. [45]

    Memformer: A memory-augmented transformer for sequence modeling

    Qingyang Wu, Zhenzhong Lan, Kun Qian, Jing Gu, Alborz Geramifard, and Zhou Yu. Memformer: A memory-augmented transformer for sequence modeling. arXiv preprint arXiv:2010.06891, 2020

    2010 arXiv

  38. [46]

    Neural fields in visual computing and beyond

    Yiheng Xie, Towaki Takikawa, Shunsuke Saito, Or Litany, Shiqin Yan, Numair Khan, Federico Tombari, James Tompkin, Vincent Sitzmann, and Srinath Sridhar. Neural fields in visual computing and beyond. In Computer Graphics Forum, volume 41, pages 641–676. Wiley Online Library, 2022

    2022

  39. [47]

    Focal self-attention for local-global interactions in vision transformers

    Jianwei Yang, Chunyuan Li, Pengchuan Zhang, Xiyang Dai, Bin Xiao, Lu Yuan, and Jianfeng Gao. Focal self-attention for local-global interactions in vision transformers. arXiv preprint arXiv:2107.00641, 2021

    2021 arXiv

  40. [48]

    mixup: Beyond empirical risk minimization

    Hongyi Zhang, Moustapha Cisse, Yann N Dauphin, and David Lopez-Paz. mixup: Beyond empirical risk minimization. arXiv preprint arXiv:1710.09412, 2017. 12

    2017 arXiv

  41. [2025]

    Retrieved from https://www.youtube.com/watch?v=xnFmnU0Pp-8

Pith tools

Reviewed August 5, 2026 · model on record in the stance chip above.