LKA: Large Kernel Adapter for Enhanced Medical Image Classification

Lu Liu; Si-Yuan Lu; Tianjin Huang; Zhe Liu; Ziquan Zhu

REVIEW 4 major objections 5 minor 31 references

LKA: Large Kernel Adapter for Enhanced Medical Image Classification

T0 review · 4 major / 5 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read A 7x7 convolutional adapter beats 11 PEFT baselines on medical images

desk verdict A clean, useful empirical recipe for medical-image PEFT, but the SOTA claim is undercut by a missing matched-kernel ConvPass control and the absence of error bars. read the letter →

arxiv 2506.19118 v4 pith:GNMD4PMC submitted 2025-06-23 cs.CE

Ziquan Zhu , Si-Yuan Lu , Tianjin Huang , Lu Liu , Zhe Liu This is my paper · ORCID

classification cs.CE

keywords Parameter-EfficientFine-TuningLargeKernelAdapterMedicalImageClassificationEffectiveReceptiveFieldChannel-wiseConvolutionTransformerTransferLearningDepthwise

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

Parameter-efficient fine-tuning (PEFT) underperforms on medical images, this paper argues, because existing adapters never enlarge the model's effective receptive field, even though medical images need long-range spatial context. The Large Kernel Adapter (LKA) addresses this by putting a channel-wise $7\times 7$ convolution inside the adapter bottleneck, and the paper reports that this small change beats 11 PEFT methods by $3.5\%$ average top-1 accuracy across five medical datasets. It also usually beats full fine-tuning while training only 0.652M parameters. The paper's contribution is a concrete mechanism, not just a new module: adapter kernel size, not parameter count, is what drives transfer performance on medical images.

What carries the argument

The central machinery is the Large Kernel Adapter (LKA), a bottleneck block defined by $x_{\text{LKA}} = W_{\text{up}}\, \text{GeLU}(\text{DWConv}_{k\times k}(W_{\text{down}}\, x)) + x$, with $W_{\text{down}}\in\mathbb{R}^{d\times \hat{d}}$, $W_{\text{up}}\in\mathbb{R}^{\hat{d}\times d}$, and $\text{DWConv}_{k\times k}$ a channel-wise (depthwise) convolution, typically $7\times 7$. The channel-wise design adds only about $(k^2+1)\hat{d}$ parameters relative to the plain adapter, keeping the receptive-field gain cheap. The paper also places LKA in parallel with the multi-head self-attention and feed-forward blocks rather than sequentially, which its ablations show is the best wiring.

What would settle it

Replace the LKA's $7\times 7$ channel-wise convolution with the same kernel whose spatial positions are randomly shuffled, keeping the training recipe unchanged; if accuracy does not drop on the five medical datasets, the large receptive field is not the active mechanism. A second check is a vanilla adapter with identical parameter count, optimizer, and schedule; if it matches LKA, the paper's central claim fails.

Watch

Extended reading notes

Core claim

The paper's central claim is that kernel size in the adapter bottleneck is the decisive variable for adapting frozen pre-trained vision models to medical imaging, and that a single channel-wise $7\times 7$ convolution is sufficient to capture the needed context. The evidence is a controlled sweep: as the kernel grows from $1\times 1$ to $7\times 7$, accuracy rises monotonically across Swin, ConvNeXt, and ViT backbones, with $7\times 7$ beating the vanilla adapter by $4.9\%$ on average. Visualizations show the effective receptive field expanding with kernel size. Additional ablations show $7\times 7$ channel-wise convolution outperforms stacked $3\times 3$ layers and dilated $3\times 3$ with dilation 3, and kernels beyond $7\times 7$ begin to hurt, giving the method a sweet spot rather than a monotonic rule.

Load-bearing premise

The load-bearing premise is that the accuracy gains come from the larger effective receptive field, not from the extra parameters, the depthwise convolution's inductive bias, or an incidental optimization effect.

Editorial extensions

If this is right

Kernel size, not parameter count, is the active ingredient: increasing the LKA kernel from $1\times1$ to $7\times7$ adds few parameters yet raises accuracy by roughly 5 points on Covid and 7 points on BUSI.
Seven-by-seven is an optimum: kernels of $9\times9$, $11\times11$, and $31\times31$ all underperform $7\times7$, so the design rule is a moderate large kernel, not an arbitrarily large one.
Parallel integration with the Transformer block (alongside MSA and FFN) is better than sequential insertion, so adapter placement matters as much as adapter structure.
A single $7\times7$ channel-wise convolution beats both a stack of three $3\times3$ convolutions and a dilated $3\times3$ kernel, indicating that one large kernel captures the global context more effectively than compounding small receptive-field expansion recipes.
LKA exceeds full fine-tuning on most datasets while updating only 0.652M parameters, so the receptive-field fix is compatible with the core promise of parameter-efficient fine-tuning.

Reading between the lines

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

The causal story rests on the correlation between measured ERF and accuracy; the paper never intervenes on the kernel's spatial structure while holding everything else fixed, so a randomized-kernel control would tell whether the receptive field or the depthwise convolution's optimization dynamics drives the gain.
Because the gain appears across three very different pre-trained architectures, the recipe may transfer to other small-sample, low-contrast imaging domains such as pathology or retinal images, but that is an extrapolation the paper does not test.
The same bottleneck may help segmentation and detection, where context is also critical, but the paper's evidence is limited to classification top-1 accuracy.
The $7\times7$ sweet spot hints at a scale-matching rule: the adapter's receptive field should match the typical spatial extent of the structures being classified; datasets with different lesion scales could shift the optimal kernel size.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Desk Editor's Note

read the letter

You should know this paper is a straightforward empirical contribution: stick a 7×7 depthwise conv inside a parallel adapter and it beats a spread of PEFT baselines on five medical datasets by a useful margin. The architecture is simple, the ablations are fairly thorough, and the code is public. The improvement over the best baseline is about 1.4 points, not the headline 3.5% (which is against the average of 11 baselines). That's still meaningful if it holds.

What's genuinely new is the targeted kernel-size sweep for adapters in medical imaging and the parallel placement result. There is prior work using convolutions in adapters — Convpass, ST-Adapter, RepAdapter — so this lives in an established design space. The paper does not hide that; it cites them. The kernel-size ablation (Table 1) and the comparison with dilated and stacked 3×3 convolutions (Table 5) give some evidence that a single large kernel is doing the work.

Now the soft spots, in proportion to how soft they are. The biggest is the missing control: LKA uses a 7×7 depthwise conv in a parallel adapter, and Convpass is a 3×3 conv in a parallel adapter. No experiment gives Convpass a 7×7 kernel. So the marginal claim that a large kernel is pivotal, rather than just a bigger kernel on an existing design, is not actually established. This is a fixable addition, but it should be done before the paper is published. Second, there are no error bars or repeated seeds anywhere. With differences of 1-2 points, that matters. Third, the kernel size, bottleneck width, and placement are selected on the same test sets that produce the final numbers, which inflates the result and makes the comparisons unfair to baselines that weren't tuned the same way. The ERF visualization is interesting but correlational; it does not prove that ERF expansion causes the accuracy gain, though that is a minor overstatement.

There are also a few citation gaps—e.g., they don't cite the large-kernel CNN literature beyond one paper—and the text has some typos ("Nnajing"). None of that changes the empirical core.

Who is this for? Anyone working on parameter-efficient fine-tuning for medical imaging. It's a credible baseline and a useful design note. The paper deserves peer review: the recipe is clear, the experiments are broad, and the issues are addressable with a validation split, repeated runs, and one matched-kernel Convpass comparison. I would not desk-reject it; I would send it back for a solid revision.

Referee Report

4 major / 5 minor

Summary. The paper proposes the Large Kernel Adapter (LKA), a parameter-efficient fine-tuning module for medical image classification. LKA consists of a down-projection, a channel-wise depthwise convolution with a 7x7 kernel, and an up-projection, inserted in parallel with the MSA and FFN modules of a frozen transformer. The authors report top-1 accuracy on five medical datasets using Swin-L, ConvNeXt-L, and ViT-L backbones, comparing against 11 PEFT baselines and full fine-tuning. They claim that a larger kernel is pivotal because it expands the effective receptive field (ERF), and they report a 3.5% average accuracy gain over the baselines. The paper also includes ablations on kernel size, integration position, bottleneck width, and alternative receptive-field-expanding convolutions.

Significance. If the claims hold, LKA is a simple, parameter-efficient recipe for adapting vision transformers to medical imaging, and the paper's kernel-size sweep is a useful empirical contribution. The breadth of the evaluation—three backbones, five datasets, and eleven baselines—is commendable, and the public code link supports reproducibility. However, the central attribution of the gain to kernel size, the state-of-the-art claim, and the statistical reliability of the reported margins are not fully established by the current experiments. The missing matched-kernel ConvPass control and the absence of repeated-seed results are the main factors limiting confidence.

major comments (4)

[Section 3.2, Table 2] The abstract and Section 3.2 state that LKA surpasses the state of the art by 3.5% in top-1 accuracy. This margin is computed against the average of the 11 PEFT baselines, not against the strongest baseline. In the average column of Table 2, the best PEFT baselines (LoRA and CIAT) reach 0.943, while LKA reaches 0.957, a gap of 1.4 percentage points. Please report the margin over the best baseline and revise the state-of-the-art claim accordingly.
[Section 3.2, Table 2] The comparison with ConvPass is confounded. LKA and ConvPass both insert a depthwise convolution into a frozen transformer, but ConvPass is evaluated only at its original 3x3 kernel size. Since the paper's central claim is that the larger kernel size is pivotal, a ConvPass variant with a 7x7 depthwise kernel is needed, together with an LKA variant using ConvPass's exact placement and module design. Without this matched-kernel control, the reported differences cannot be attributed specifically to the kernel size rather than to placement, activation, or the presence of the down/up projections.
[Sections 3.1 and 3.3] Hyperparameters are selected on the same test split that produces the final accuracies. The kernel size (Table 3), bottleneck width (Table 6), and integration position (Table 4) are all chosen using the 80/20 test split described in Section 3.1, and all tables report a single run with no error bars. Please add a validation split for model selection and report the mean and standard deviation over at least three seeds. This is essential for supporting the small 1.4-point margin over the best baseline.
[Section 3.2, item 2 and Figure 2] The ERF analysis is presented only as visualizations; no quantitative ERF metric is reported, and no ERF estimate is given for the ConvPass baseline. The statement that ERF expansion is the causal mechanism behind the accuracy gain is therefore an inference from correlation. Please provide a quantitative ERF measure (for example, the standard deviation or effective area of the contribution map) for each configuration, and distinguish the kernel-size sweep as the only intervention that directly supports the causal claim.

minor comments (5)

[Equation (2)] The residual term in Eq. (2) is dimensionally inconsistent: the term 'x' is added inside the parentheses before the up-projection, but 'x' has dimension d while 'W_down * x' has dimension d_hat. The intended formula is likely 'x_LKA = W_up * GeLU(DWConv(W_down * x)) + x' or an equivalent form; please correct it.
[Section 3.3, parameter-efficiency paragraph] The total parameter count is stated as '2*d*d_hat + (k^2+2)*d_hat + d', but the text then says LKA adds only '(k^2+1)*d_hat' parameters over the vanilla adapter (which has '2*d*d_hat + d_hat + d'). These expressions differ by one bias term; please make the count consistent.
[Table 2] The column header 'Kernel Size' is meaningful only for LKA; the baseline rows should leave this cell empty or report each baseline's convolutional kernel when applicable, because currently the combined header implies a quantity that is not defined for most baselines.
[Section 3.1] Dataset details are incomplete: please report the number of images per class, input resolution, and the random split procedure, and correct the typo 'Nnajing' in the affiliations.
[Section 3.1] There is a doubled article in 'we report the the top-1 accuracy'; also, Section 3.2 item 1 states the 4.9% margin without noting that it is in percentage points against the 'None' (vanilla adapter) row of Table 1.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: all central claims are supported by direct experiments and external baselines, with no fitted quantity relabeled as a prediction.

full rationale

LKA is an architectural proposal, and the paper's claims are empirical: kernel-size sweeps (Table 1 and Table 3), placement ablations (Table 4), receptive-field-recipe comparisons (Table 5), and parameter-count controls (Figure 3) directly compare test accuracy of the proposed module against vanilla adapters and eleven named baselines. There is no equation in which an output is defined in terms of the claimed conclusion, no parameter fitted to a subset and then 'predicted' on a closely related quantity, and no load-bearing self-citation: the cited prior works (ChannelNets, ConvPass, RepMLPNet, etc.) are external to the present authors and are used as sources of standard components or comparison methods. The ERF visualization is a measurement rather than a derivation, and the causal attribution of accuracy gains to ERF expansion is not fully isolated — for instance, Table 2 evaluates ConvPass only at its original 3x3 kernel, so no matched-kernel ConvPass-7x7 control is reported — but this is an experimental-control issue, not a circular reduction. The core superiority claim is therefore self-contained against the reported benchmarks, and the circularity score is 0.

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

The paper relies on standard pretrained backbones and an empirical evaluation protocol. No first-principles derivation is attempted. The key free choices are architecture hyperparameters selected on the test sets. No new physical or conceptual entities are introduced.

free parameters (3)

Kernel size k = 7x7
Chosen because Table 3 on Swin-T shows 7x7 gives the best average top-1 accuracy over the five test sets; all reported LKA results use this value.
Bottleneck width d_hat = 8
Selected from Table 6 as the best width on the same five test sets used for the final results.
Adapter integration position = parallel with MSA and FFN
Position (a) in Table 4 is selected using the same test sets; the final results use this placement.

assumptions (3)

domain assumption Pretrained backbones frozen during fine-tuning still provide representations useful for medical image classification.
Invoked in Section 3.1, where all backbones are frozen and only LKA weights are updated; if false, the PEFT framing collapses.
domain assumption The 80/20 single split and top-1 accuracy are sufficient to rank methods reliably.
Section 3.1 states an 80/20 split with no cross-validation or repeated seeds; the 3.5% headline depends on this assumption.
domain assumption The effective receptive field (ERF) measurement with 50 resized validation images reflects the spatial context used for classification.
Section 3.2 item 2 uses this to support the mechanism claim; the link from ERF to accuracy is not proven.

how reviews work

0 comments

Cite this review

Pith. "Pith review of LKA: Large Kernel Adapter for Enhanced Medical Image Classification." pith.science (2026). https://pith.science/paper/GNMD4PMC

@misc{pith2026250619118,
  author       = {Pith},
  title        = {Pith review of: LKA: Large Kernel Adapter for Enhanced Medical Image Classification},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/GNMD4PMC}},
  note         = {Machine review of arXiv:2506.19118}
}

read the original abstract

Despite the notable success of current Parameter-Efficient Fine-Tuning (PEFT) methods across various domains, their effectiveness on medical datasets falls short of expectations. This limitation arises from two key factors: (1) medical images exhibit extensive anatomical variation and low contrast, necessitating a large receptive field to capture critical features, and (2) existing PEFT methods do not explicitly address the enhancement of receptive fields. To overcome these challenges, we propose the Large Kernel Adapter (LKA), designed to expand the receptive field while maintaining parameter efficiency. The proposed LKA consists of three key components: down-projection, channel-wise large kernel convolution, and up-projection. Through extensive experiments on various datasets and pre-trained models, we demonstrate that the incorporation of a larger kernel size is pivotal in enhancing the adaptation of pre-trained models for medical image analysis. Our proposed LKA outperforms 11 commonly used PEFT methods, surpassing the state-of-the-art by 3.5% in top-1 accuracy across five medical datasets.

Figures

Figures reproduced from arXiv: 2506.19118 by the authors.

**Figure 1.** The structures of Adapter(a) and LKA(b), full fine-tuning (c) and LKA-tuning (d). During training, only LKAs are trainable, while all other layers remain frozen. 3 Experiments and Results 3.1 Experiment Settings To demonstrate the effectiveness of our proposed LKA method, we conduct experiments using 3 popular pre-trained models across 5 medical datasets and report the the top-1 accuracy in all experiments. Additio… view at source ↗

**Figure 2.** Effective receptive fields (ERFs) for LKA with various kernel sizes based on pretrained Swin-T. 0.32M 0.34M 0.36M 0.38M #Trainable Parameters 92 94 A C C(%)( ) 1×1 3×3 5×5 7×7 8 9 10 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 LKA Vanilla Adapter (a) Covid 0.32M 0.34M 0.36M 0.38M #Trainable Parameters 80.0 82.5 85.0 87.5 A C C(%)( ) 1×1 3×3 5×5 7×7 8 9 10 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 LKA Vani… view at source ↗

**Figure 3.** Comparison of LKA and Vanilla Adapter with varying #Trainable Parameters based on pre-trained Swin-T and two datasets: Covid (a) and BUSI (b). ❸ Large Kernel Matters Instead of #Trainable Parameters. When increasing the kernel size in convolutional layers, it inevitably leads to an increase in the number of trainable parameters. To substantiate our claim that large kernel convolutions are critical for effectively a… view at source ↗

Review history (2 revisions) →

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

31 extracted references · 19 canonical work pages

[1]

ACM Computing Surveys55(9), pp

Liu, P., Yuan, W., Fu, J., Jiang, Z., Hayashi, H., Neubig, G.: Pre-train, prompt, and predict: A systematic survey of prompting methods in natural language processing. ACM Computing Surveys55(9), pp. 1–35 (2023). ACM New York, NY

work page 2023
[2]

BMC Medical Imaging 22(1), pp

Kim, H.E., Cosa-Linan, A., Santhanam, N., Jannesari, M., Maros, M.E., Ganslandt, T.: Transfer learning for medical image classification: A literature review. BMC Medical Imaging 22(1), pp. 69 (2022). Springer

work page 2022
[3]

arXiv preprint arXiv:2006.16668 (2020)

Lepikhin, D., Lee, H., Xu, Y., Chen, D., Firat, O., Huang, Y., Krikun, M., Shazeer, N., Chen, Z.: Gshard: Scaling giant models with conditional computation and au- tomatic sharding. arXiv preprint arXiv:2006.16668 (2020)

arXiv 2020
[4]

European Conference on Computer Vision, pp

Jia, M., Tang, L., Chen, B.-C., Cardie, C., Belongie, S., Hariharan, B., Lim, S.- N.: Visual prompt tuning. European Conference on Computer Vision, pp. 709–727 (2022). Springer

work page 2022
[5]

International Conference on Machine Learning, pp

Houlsby, N., Giurgiu, A., Jastrzebski, S., Morrone, B., De Laroussilhe, Q., Ges- mundo, A., Attariyan, M., Gelly, S.: Parameter-efficient transfer learning for NLP. International Conference on Machine Learning, pp. 2790–2799 (2019). PMLR

work page 2019
[6]

arXiv preprint arXiv:2106.09685 (202

Hu, E.J., Shen, Y., Wallis, P., Allen-Zhu, Z., Li, Y., Wang, S., Chen, W.: LoRA: Low-rank adaptation of large language models. arXiv preprint arXiv:2106.09685 (202

arXiv
[7]

Cognitive Computation, pp

Li, H., Nan, Y., Del Ser, J., Yang, G.: Large-kernel attention for 3D medical image segmentation. Cognitive Computation, pp. 1–15 (2023). Springer. 10 Z.Zhu and S.Lu, et al

work page 2023
[8]

Advances in Neural Information Processing Systems 31 (2018)

Gao, H., Wang, Z., Ji, S.: ChannelNets: Compact and efficient convolutional neural networks via channel-wise convolutions. Advances in Neural Information Processing Systems 31 (2018)

work page 2018

Show all 31 references

[9]

Proceedings of the IEEE/CVF International Conference on Computer Vision, pp

Liu, Z., Lin, Y., Cao, Y., Hu, H., Wei, Y., Zhang, Z., Lin, S., Guo, B.: Swin Trans- former: Hierarchical vision transformer using shifted windows. Proceedings of the IEEE/CVF International Conference on Computer Vision, pp. 10012–10022 (2021)

2021
[10]

Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp

Liu, Z., Mao, H., Wu, C.-Y., Feichtenhofer, C., Darrell, T., Xie, S.: A ConvNet for the 2020s. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 11976–11986 (2022)

2022
[11]

arXiv preprint arXiv:2010.11929 (2020)

Dosovitskiy, A., Beyer, L., Kolesnikov, A., Weissenborn, D., Zhai, X., Unterthiner, T., Dehghani, M., Minderer, M., Heigold, G., Gelly, S., et al.: An image is worth 16x16 words: Transformers for image recognition at scale. arXiv preprint arXiv:2010.11929 (2020)

2020 arXiv
[12]

arXiv preprint arXiv:2207.03620 (2022)

Liu, S., Chen, T., Chen, X., Chen, X., Xiao, Q., Wu, B., Kärkkäinen, T., Pech- enizkiy, M., Mocanu, D., Wang, Z.: More ConvNets in the 2020s: Scaling up kernels beyond 51×51 using sparsity. arXiv preprint arXiv:2207.03620 (2022)

2022 arXiv
[13]

Advances in Neural Information Processing Systems35, pp

Pan, J., Lin, Z., Zhu, X., Shao, J., Li, H.: ST-Adapter: Parameter-efficient image- to-video transfer learning. Advances in Neural Information Processing Systems35, pp. 26462–26477 (2022)

2022
[14]

Yang, T., Zhu, Y., Xie, Y., Zhang, A., Chen, C., Li, M.: AIM: Adapting image modelsforefficientvideoactionrecognition.arXivpreprintarXiv:2302.03024(2023)

2023 arXiv
[15]

arXiv preprint arXiv:1909.08478 (2019)

Bapna, A., Arivazhagan, N., Firat, O.: Simple, scalable adaptation for neural ma- chine translation. arXiv preprint arXiv:1909.08478 (2019)

2019 arXiv
[16]

arXiv preprint arXiv:2207.07039 (2022)

Jie, S., Deng, Z.-H.: Convolutional bypasses are better vision transformer adapters. arXiv preprint arXiv:2207.07039 (2022)

2022 arXiv
[17]

Biomedical Signal Processing and Control71, pp

Girdhar, A., Kapur, H., Kumar, V.: Classification of white blood cell using con- volution neural network. Biomedical Signal Processing and Control71, pp. 103156 (2022). Elsevier

2022
[18]

Data in Brief28, pp

Al-Dhabyani, W., Gomaa, M., Khaled, H., Fahmy, A.: Dataset of breast ultrasound images. Data in Brief28, pp. 104863 (2020). Elsevier

2020
[19]

Journal of Neurology, Neurosurgery & Psychiatry74(3), pp

Summers,D.:HarvardWholeBrainAtlas:www.med.harvard.edu/AANLIB/home.html. Journal of Neurology, Neurosurgery & Psychiatry74(3), pp. 288–288 (2003). BMJ Publishing Group Ltd

2003
[20]

IEEE Access 8, pp

Rahman, T., Khandakar, A., Kadir, M.A., Islam, K.R., Islam, K.F., Mazhar, R., Hamid, T., Islam, M.T., Kashem, S., Mahbub, Z.B., et al.: Reliable tuberculosis detection using chest X-ray with deep learning, segmentation and visualization. IEEE Access 8, pp. 191586–191601 (2020). IEEE

2020
[21]

NPJ Digital Medicine6(1) (2023)

Choy, S.P., et al.: Systematic review of deep learning image analyses for the diag- nosis and monitoring of skin disease. NPJ Digital Medicine6(1) (2023)

2023
[22]

Advances in Neural Information Processing Systems35, pp

Liang, W., Yuan, Y., Ding, H., Luo, X., Lin, W., Jia, D., Zhang, Z., Zhang, C., Hu, H.: Expediting large-scale vision transformer for dense prediction without fine- tuning. Advances in Neural Information Processing Systems35, pp. 35462–35477 (2022)

2022
[23]

Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp

Sandler, M., Zhmoginov, A., Vladymyrov, M., Jackson, A.: Fine-tuning image transformers using learnable memory. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 12155–12164 (2022)

2022
[24]

arXiv preprint arXiv:2104.08154 (2021)

Zhu, Y., Feng, J., Zhao, C., Wang, M., Li, L.: Counter-interference adapter for multilingual machine translation. arXiv preprint arXiv:2104.08154 (2021)

2021 arXiv
[25]

arXiv preprint arXiv:2304.06600 (2023)

Sharma, M., Fantacci, C., Zhou, Y., Koppula, S., Heess, N., Scholz, J., Aytar, Y.: Lossless adaptation of pretrained vision models for robotic manipulation. arXiv preprint arXiv:2304.06600 (2023). LKA: Large Kernel Adapter for Enhanced Medical Image Classification 11

2023 arXiv
[26]

arXiv preprint arXiv:2302.08106 (2023)

Luo, G., Huang, M., Zhou, Y., Sun, X., Jiang, G., Wang, Z., Ji, R.: To- wards efficient visual adaptation via structural re-parameterization. arXiv preprint arXiv:2302.08106 (2023)

2023 arXiv
[27]

Advances in Neural Information Processing Systems35, pp

Chen, S., Ge, C., Tong, Z., Wang, J., Song, Y., Wang, J., Luo, P.: AdaptFormer: Adapting vision transformers for scalable visual recognition. Advances in Neural Information Processing Systems35, pp. 16664–16678 (2022)

2022
[28]

arXiv preprint arXiv:2106.10199 (2021)

Zaken, E.B., Ravfogel, S., Goldberg, Y.: BitFit: Simple parameter-efficient fine-tuning for transformer-based masked language models. arXiv preprint arXiv:2106.10199 (2021)

2021
[29]

Distill 4(11), pp

Araujo, A., Norris, W., Sim, J.: Computing receptive fields of convolutional neural networks. Distill 4(11), pp. e21 (2019)

2019
[30]

Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp

Ding, X., Chen, H., Zhang, X., Han, J., Ding, G.: RepMLPNet: Hierarchical vision MLP with re-parameterized locality. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 578–587 (2022)

2022
[31]

arXiv preprint arXiv:1511.07122 (2015)

Yu, F., Koltun, V.: Multi-scale context aggregation by dilated convolutions. arXiv preprint arXiv:1511.07122 (2015)

2015 arXiv

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.

[1] [1]

ACM Computing Surveys55(9), pp

Liu, P., Yuan, W., Fu, J., Jiang, Z., Hayashi, H., Neubig, G.: Pre-train, prompt, and predict: A systematic survey of prompting methods in natural language processing. ACM Computing Surveys55(9), pp. 1–35 (2023). ACM New York, NY

work page 2023

[2] [2]

BMC Medical Imaging 22(1), pp

Kim, H.E., Cosa-Linan, A., Santhanam, N., Jannesari, M., Maros, M.E., Ganslandt, T.: Transfer learning for medical image classification: A literature review. BMC Medical Imaging 22(1), pp. 69 (2022). Springer

work page 2022

[3] [3]

arXiv preprint arXiv:2006.16668 (2020)

Lepikhin, D., Lee, H., Xu, Y., Chen, D., Firat, O., Huang, Y., Krikun, M., Shazeer, N., Chen, Z.: Gshard: Scaling giant models with conditional computation and au- tomatic sharding. arXiv preprint arXiv:2006.16668 (2020)

arXiv 2020

[4] [4]

European Conference on Computer Vision, pp

Jia, M., Tang, L., Chen, B.-C., Cardie, C., Belongie, S., Hariharan, B., Lim, S.- N.: Visual prompt tuning. European Conference on Computer Vision, pp. 709–727 (2022). Springer

work page 2022

[5] [5]

International Conference on Machine Learning, pp

Houlsby, N., Giurgiu, A., Jastrzebski, S., Morrone, B., De Laroussilhe, Q., Ges- mundo, A., Attariyan, M., Gelly, S.: Parameter-efficient transfer learning for NLP. International Conference on Machine Learning, pp. 2790–2799 (2019). PMLR

work page 2019

[6] [6]

arXiv preprint arXiv:2106.09685 (202

Hu, E.J., Shen, Y., Wallis, P., Allen-Zhu, Z., Li, Y., Wang, S., Chen, W.: LoRA: Low-rank adaptation of large language models. arXiv preprint arXiv:2106.09685 (202

arXiv

[7] [7]

Cognitive Computation, pp

Li, H., Nan, Y., Del Ser, J., Yang, G.: Large-kernel attention for 3D medical image segmentation. Cognitive Computation, pp. 1–15 (2023). Springer. 10 Z.Zhu and S.Lu, et al

work page 2023

[8] [8]

Advances in Neural Information Processing Systems 31 (2018)

Gao, H., Wang, Z., Ji, S.: ChannelNets: Compact and efficient convolutional neural networks via channel-wise convolutions. Advances in Neural Information Processing Systems 31 (2018)

work page 2018