Contrastive meta-domain adaptation for robust skin lesion classification across clinical and acquisition conditions

Emanoel dos Santos; Erico Medeiros; F\'abio Papais; Francisco Filho; Kelvin Cunha; Paulo Borba; Rodrigo Mota; Thales Bezerra; Tsang Ing Ren

REVIEW 5 major objections 4 minor 16 references

Contrastive meta-domain adaptation for robust skin lesion classification across clinical and acquisition conditions

T0 review · 5 major / 4 minor · reviewed 2026-08-02 · deepseek-v4-flash

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

Pith's one-line read This paper claims a two-stage training pipeline can adapt dermoscopy-trained skin-lesion classifiers to clinical smartphone photos while preserving prior-domain knowledge.

desk verdict Plausible domain-adaptation recipe undermined by an unexplained DDI baseline collapse and an unspecified label mapping across datasets; the evidence needs a re-run before the headline results are credible. read the letter →

arxiv 2602.19857 v2 pith:IGK3J4YY submitted 2026-02-23 cs.CV

Rodrigo Mota , Kelvin Cunha , Emanoel dos Santos , Fábio Papais , Francisco Filho , Thales Bezerra , Erico Medeiros , Paulo Borba

show 1 more author

Tsang Ing Ren

This is my paper · ORCID

classification cs.CV

keywords skinlesionclassificationdomainadaptationcontrastivelearningcontinualmeta-domaindermoscopyclinicalimagingcatastrophicforgetting

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 is trying to establish that the poor performance of skin-lesion classifiers on clinical images is caused by acquisition artifacts and domain shift, and that both can be mitigated with a domain-aware training recipe. The recipe has two parts: a contrastive pre-training step that makes the model's representations invariant to image degradations and separates different lesions, and a guided-tuning step that adapts the model to the clinical domain using a small calibration subset while keeping what it learned on dermoscopy. On two clinical smartphone datasets, the combined pipeline raises accuracy/F1 from very low naive baselines (0.35/0.38 and 0.12/0.05) to 0.88/0.84 and 0.79/0.81, and it preserves performance on the previously learned dermoscopic domain. A sympathetic reader would care because this addresses the main barrier to deploying dermatology AI in real clinics: models trained on clean dermoscopic datasets fail when faced with variable clinical photos.

What carries the argument

The central object is the visual meta-domain: a synthetic variant of the source dataset whose appearance is transformed to match the target clinical domain using color statistics (LAB mean and standard deviation) and blur statistics (Laplacian variance and gradient measures) estimated from a small calibration subset of target images. K such meta-domains are generated stochastically and used in a guided-tuning loss that combines the target-domain objective with meta-domain losses, preventing catastrophic forgetting while adapting. The companion machinery is multi-transform contrastive pre-training, which uses a multi-positive InfoNCE loss on stochastically augmented views of dermoscopic image

What would settle it

Take a clinical test set acquired at a site that contributed no images to the calibration subset and compare per-class accuracy of guided-tuning against naive training; if the accuracy reverts to naive levels, the calibration statistics do not capture the full domain shift and the central claim fails.

Watch

Extended reading notes

Core claim

The central claim is that adaptation to clinical skin-lesion images can be made stable—accurate on the new domain and non-forgetting of the old one—by combining contrastive pre-training with guided tuning. Guided tuning works by extracting a small calibration subset from the target clinical data, estimating its global color statistics (LAB mean/std) and blur characteristics (Laplacian variance, gradient measures), and using those estimates to transform source dermoscopic images into K synthetic meta-domains. The model is then optimized with a loss that mixes the target-domain fine-tuning objective with evaluations on these meta-domains, which acts as a continual-learning regularizer. In the

Load-bearing premise

The pipeline assumes the dermoscopic source and clinical target share the same lesion classes, and that color and blur statistics measured on a small calibration subset of the target training set represent the entire clinical domain.

Editorial extensions

If this is right

Contrastive pre-training on dermoscopic images yields models that degrade less under blur, noise, illumination shifts, and overexposure than models trained with standard supervised backpropagation.
Guided tuning improves clinical dataset accuracy/F1 over naive training, standard fine-tuning, and random augmentation, and it does so even when the target-domain training set is small.
The guided-tuning loss preserves performance on previously learned domains, so a model can be continually adapted to new clinical sites without catastrophic forgetting.
Combining contrastive pre-training with guided tuning gives the best reported results on both clinical datasets evaluated, suggesting the two components are complementary.

Reading between the lines

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

If the mechanism is primarily matching low-level color and blur statistics, the same guided-tuning recipe may transfer to other medical imaging domains where acquisition shifts dominate over content differences, but that transfer is an extension the paper does not test.
The protocol's reliance on shared label spaces is untested; on a clinical dataset with a different lesion taxonomy, the adaptation objective would need explicit label mapping or an open-set formulation before those accuracy numbers are meaningful.
A direct robustness test would vary the size and acquisition-site composition of the calibration subset; stable performance under such variation would support the claim that the estimated statistics capture the domain shift rather than overfit the calibration images.
The multi-transform contrastive objective could be applied to unlabeled clinical images to reduce domain shift further, a semi-supervised extension implied by the method's design but not explored here.

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

Colleague,

Quick read of arXiv:2602.19857. The paper proposes a two-stage recipe: multi-transform InfoNCE contrastive pre-training on HAM10000, then a “guided-tuning” meta-domain adaptation that simulates K=2 meta-domains by transferring LAB color and blur statistics from a small calibration subset of the target. That is a reasonable engineering idea, and the combination on HAM→PAD/DDI appears to be new. If the numbers hold, it is a useful recipe for deploying classifiers on smartphone dermatology images. The motivation is sound and the method description is mostly clear; credit where due for reusing Volpi-style meta-learning with a fixed loss objective and target-guided augmentations.

The problem is the empirical support. The central table has no seeds, no error bars, no confidence intervals. More seriously, the Naive backprop baseline on DDI collapses to ACC 0.12, F1 0.05, far below chance for any usual number of classes and far below the cited Daneshjou baseline of 0.77. That is not a plausible lower bound; it smells like a label-alignment bug between HAM and DDI classes. The paper never states the label mapping across HAM10000, PAD-UFES-20, and DDI. If the taxonomies differ, Eq. (4) and Table 1 are not well-defined, and the Fig. 5 “previously learned domain” curves don’t mean what the authors claim. Because every reported gain is relative to that collapsed baseline, the headline result could be an artifact. The stress-test note lands.

There are also smaller issues: hyperparameters (β1=β2=0.5, K=2, augmentation probability, temperature τ) are hand-set without sensitivity analysis, and no code or trained models are provided. These are fixable, but they matter for reproducibility.

Who this is for: people working on medical domain adaptation and deployment of dermatology classifiers. The paper could be a solid engineering contribution after careful revision, but not as is. I would not desk-reject it: the method is coherent and worth a serious referee’s time, but only with a strict request to provide code, variance estimates, the exact label mapping, and an explanation—or fix—of the DDI Naive baseline. If those don’t hold, the paper should not be published.

Referee Report

5 major / 4 minor

Summary. The paper introduces a two-stage domain-adaptation pipeline for skin-lesion classification: a multi-transform contrastive pre-training step on dermoscopic images (HAM10000) followed by a 'guided-tuning' (GT) meta-domain adaptation stage that uses target-domain color and blur statistics to simulate K=2 meta-domains and fine-tune the model on clinical datasets (PAD-UFES-20 and DDI). The authors report that CT-pretrain + GT achieves the best accuracy and F1 scores on both clinical datasets, while reducing catastrophic forgetting of previously learned dermoscopic domains. The central claim is that the proposed guided-tuning strategy enables stable adaptation to clinical domains while preserving knowledge from dermoscopic training.

Significance. If substantiated, the proposed pipeline would be a practical contribution to clinical skin-lesion classification, where domain shift is a recognized barrier to deployment. The idea of using target-domain color and blur statistics to synthesize 'meta-domains' is interesting and potentially useful. However, the experimental validation as presented is not reliable: the paper never defines a shared label space across datasets, Table 1 reports no variance estimates or seeds, and the DDI Naive baseline (ACC 0.12, F1 0.05) is implausibly low, suggesting a label-alignment bug. Because all headline gains are relative to these baselines, the central claim is currently unsupported. The paper is also difficult to reproduce due to missing implementation details (optimizer, epochs, batch size, augmentation magnitudes). These issues are load-bearing rather than cosmetic.

major comments (5)

[§3.3, Table 1, Eq. (4)] The paper never defines the label mapping between HAM10000's seven classes and the taxonomies of PAD-UFES-20 and DDI. Eq. (4) uses the cross-entropy loss L_o on both source and target sets; this loss is undefined if the label spaces differ. If the output head is replaced or reinitialized during adaptation, the claim of 'preserving previously learned domains' (Fig. 5) is not meaningful without specifying how the source labels are mapped. The DDI Naive baseline (ACC 0.12, F1 0.05) is far below chance and far below the cited Daneshjou et al. baseline of 0.77, strongly suggesting a label-index mismatch. Since every improvement in Table 1 is measured against this Naive baseline, the central experimental claim is not currently supported.
[Table 1, Figs. 3–5] No variance estimates, number of seeds, or confidence intervals are reported anywhere. All results are single runs. Differences such as GT vs. CT+GT on DDI (ACC 0.79 vs 0.79, F1 0.79 vs 0.81) are within plausible run-to-run noise. Without at least five seeds with mean±std and statistical testing, the reported improvements cannot be interpreted. This is especially important for a domain-adaptation method, where training is stochastic and the calibration subsets are randomly partitioned.
[§3.2.2, Eq. (4)] The calibration subset size is not specified, yet the method's domain simulation relies on estimating LAB color statistics and blur features from this subset. The manuscript does not characterize how sensitive the estimates are to subset size or composition. In addition, the hyperparameters K=2, β1=β2=0.5, and augmentation probability 0.5 are hand-set with no sensitivity analysis. The claim that GT enables 'stable adaptation with limited target-domain data' (Fig. 4) is based on a single curve without error bars, so it does not establish stability.
[§4.1, Fig. 3, Eq. (2)] The degraded test set used to evaluate robustness is not described reproducibly: the types and magnitudes of blur, sensor noise, illumination shifts, motion blur, and overexposure are not specified. Moreover, the number of augmented views N and the temperature τ in Eq. (2) are never reported. These details are essential for reproducing the contrastive pre-training experiments and for assessing the claim that CT-pretrain improves robustness to clinical artifacts.
[§3.3, experimental setup] The manuscript omits basic training details: the specific EfficientNet variant, image resolution, optimizer, learning rate schedule, batch size, number of epochs, and hardware. It also does not state whether the HAM10000-trained model is fine-tuned as a whole or only the head. These omissions make the experiments non-reproducible and prevent an independent check of the reported numbers.

minor comments (4)

[Throughout] Numerous typographical errors, including 'ADAPTA TION' and 'Clasifica tion' in the title, 'Adressing' in §3.2, 'Constrative' in §4.1, and 'CT-pretain' in Table 1. The manuscript should be proofread.
[Eqs. (1)–(2)] The relation between the single-positive InfoNCE loss in Eq. (1) and the multi-positive loss in Eq. (2) is not clearly explained. The notation α=1/(N+1) is confusing because the sum runs over k=0..N. Please clarify the indexing and the exact composition of both losses.
[§3.2.2] The set S_adapt = S_cal ∪ S_ds′ is not precisely defined. In particular, it is unclear whether S_ds′ is the entire transformed source set or a sampled mini-batch, and how the union is formed when the calibration subset and the source set have different sizes and label spaces.
[Fig. 4] The x-axis is labeled 'Training set size' with values 0%–100%, but the text says 'PAD validation increasing training set size'. It would be clearer to state explicitly that the percentages refer to the fraction of the PAD-UFES-20 training set used and how the remaining fraction is treated.

Circularity Check

0 steps flagged · score 1.0 of 10

No significant circularity: guided-tuning is standard target-calibrated domain adaptation; the only self-citation is non-load-bearing.

full rationale

The paper's derivation chain is empirical and self-contained. The contrastive losses (Eqs. 1–2) are standard InfoNCE objectives applied during pre-training on HAM10000, and the guided-tuning objective (Eq. 4) fine-tunes the model on target training/calibration data (S_cal ⊆ S_dt) with color and blur statistics estimated from that same target subset. This is legitimate supervised domain adaptation, not a prediction from a fitted parameter: the reported numbers are evaluated on held-out test splits, and no metric is a renamed fitted value. The only author-overlapping citation is reference [3] (Emanoel Santos et al.), which supports the generic claim that clinical artifacts degrade model performance; it is not load-bearing for the method or for the headline gains in Table 1. The concerns about undefined label-space compatibility between HAM10000, PAD-UFES-20, and DDI, and the below-chance DDI Naive baseline (ACC 0.12), are correctness/protocol risks rather than circular reductions of the central claim. No self-definitional equality, fitted-input-called-prediction step, or self-citation chain forcing the result was found.

Assumptions & free parameters 6 free parameters · 4 assumptions · 1 invented entities

The paper's central claim rests on several unstated modeling choices: contrastive pretraining is assumed to transfer, the calibration statistics are assumed to capture the target domain, and the three datasets are assumed to share compatible labels. These are domain assumptions rather than proven facts, and the hand-set hyperparameters are not ablated.

free parameters (6)

contrastive temperature τ = not reported
Appears in Eqs. 1-2 and controls the sharpness of the contrastive distribution; no value or tuning procedure is given.
adaptation weights β1, β2 = 0.5, 0.5
Eq. 4 weights the meta-update terms equally; chosen by hand with no sensitivity analysis.
number of meta-domains K = 2
The calibration set is randomly partitioned into K=2 meta-domains (§3.2.2); no ablation over K is reported.
augmentation probability = 0.5
All Albumentations transforms are applied stochastically with probability 0.5.
number of augmented views N = not reported
The multi-transform contrastive loss uses N stochastic augmentations per image, but N is never specified.
calibration subset size = not reported
A calibration subset Scal ⊆ Sdt is used to estimate domain statistics; its size is not given, affecting the reliability of color/blur estimates.

assumptions (4)

domain assumption InfoNCE/contrastive pre-training produces invariant features that transfer to downstream skin-lesion classification.
Eqs. 1-2 are assumed to yield robustness to clinical variability, but no analysis or ablation isolates this mechanism from the rest of the pipeline.
ad hoc to paper LAB color statistics and blur metrics from a small target calibration subset are sufficient to simulate the target domain via Albumentations.
§3.2.2 selects color transfer and blur degradations heuristically; no validation shows these transforms cover the true domain shift.
domain assumption HAM10000, PAD-UFES-20, and DDI share compatible label spaces, allowing cross-dataset fine-tuning and evaluation without remapping.
§3.3 and Table 1 evaluate transfer across the three datasets but never describe how classes are aligned or whether the taxonomies match.
domain assumption Gradient-based optimization with Eq. 4 approximates meta-learning and reduces catastrophic forgetting.
The guided-tuning update is motivated by continual learning, but the claim that it preserves prior domains relies on Figure 5, which is not shown in the text and has no quantitative analysis.

invented entities (1)

visual meta-domain
purpose: A partition of the target calibration data into appearance clusters that guide the adaptation of source-domain representations.
This is a new methodological construct introduced in §3.2.2 and Figure 2. It has no external falsifiable handle; it is an internal convenience for organizing target appearance statistics.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Contrastive meta-domain adaptation for robust skin lesion classification across clinical and acquisition conditions." pith.science (2026). https://pith.science/paper/IGK3J4YY

@misc{pith2026260219857,
  author       = {Pith},
  title        = {Pith review of: Contrastive meta-domain adaptation for robust skin lesion classification across clinical and acquisition conditions},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/IGK3J4YY}},
  note         = {Machine review of arXiv:2602.19857}
}

read the original abstract

Deep learning models for dermatological image analysis remain sensitive to acquisition variability and domain-specific visual characteristics, leading to performance degradation when deployed in clinical settings. We investigate how visual artifacts and domain shifts affect deep learning-based skin lesion classification. We propose an adaptation strategy, grounded in the idea of visual meta-domains, that transfers visual representations from larger dermoscopic datasets into clinical image domains, thereby improving generalization robustness. Experiments across multiple dermatology datasets show consistent gains in classification performance and reduced gaps between dermoscopic and clinical images. These results emphasize the importance of domain-aware training for deployable systems.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

16 extracted references · 2 linked inside Pith

[1]

This gap remains largely from biases in image acquisition and dataset composition, which shape the visual features learned during training and can lead to unreliable predictions

INTRODUCTION Deep learning has shown strong potential for supporting skin cancer diagnosis [1], yet current models often fail to generalize in practical clinical settings [2]. This gap remains largely from biases in image acquisition and dataset composition, which shape the visual features learned during training and can lead to unreliable predictions. Al...
[2]

Ideally, f(x)should emulate the human decision,H(x)→y

PROBLEM FORMULA TION In supervised learning, we aim to approximate a functionf(x)that predicts a labelˆyfrom a labeled setS={(x i, yi)}N i=1. Ideally, f(x)should emulate the human decision,H(x)→y. However, since the true data distributionp(x, y)is unknown, the sample setS may contain biases or noise, which limits how closelyfcan approxi- mateH. We express...
[3]

Motivation Despite advances in vision algorithms, adoption in dermatological practice remains limited due to variability in imaging conditions

METHODOLOGY 3.1. Motivation Despite advances in vision algorithms, adoption in dermatological practice remains limited due to variability in imaging conditions. These challenges arise from task-related noise inherent to clinical workflows. In this study, we analyze how these factors affect model decisions and introduce a domain-adaptation pipeline designe...

arXiv 2026
[4]

These models were optimized to classify in dermoscopic images

EXPERIMENTS AND DISCUSSIONS We trained the model on the HAM10000 [8], a large collection captured with magnifying devices that highlight fine lesion details. These models were optimized to classify in dermoscopic images. However, our goal is to apply them in clinical scenarios, using datasets such as PAD-UFES-20 [5] and DDI [6]. These datasets are compara...
[5]

Traditional machine learning approaches often struggle with domain variations, which critically impact real-world reliabil- ity

CONCLUSION In this work, we introduce a training strategy to mitigate perfor- mance degradation in clinical models for dermatological lesion as- sessment. Traditional machine learning approaches often struggle with domain variations, which critically impact real-world reliabil- ity. To address this, we propose a contrastive pre-training strat- egy that en...
[6]

Ethical approval was not required, as confirmed by the license attached to the open- access data

COMPLIANCE WITH ETHICAL STANDARDS This research study was conducted retrospectively using human sub- ject data made available in open access [8, 5, 6]. Ethical approval was not required, as confirmed by the license attached to the open- access data
[7]

ACKNOWLEDGMENTS This work was partially supported by INES.IA (National Institute of Science and Technology for Software Engineering Based on and for Artificial Intelligence) www.ines.org.br, CNPq grant 408817/2024-

2024
[8]

8,248, Table 1

The project was supported by the Ministry of Science, Technol- ogy, and Innovation of Brazil, with resources from Law No. 8,248, Table 1. Comparison of naive training, traditional fine-tuning with augmentation, and the proposed contrastive pre-training (CT-pretrain) and guided-tuning (GT) strategies. We replicate the baseline methods reported in [5, 6] an...

arXiv 1991

Show all 16 references

[9]

Machine learning in dermatology: cur- rent applications, opportunities, and limitations,

Stephanie Chan et al., “Machine learning in dermatology: cur- rent applications, opportunities, and limitations,”Dermatology and therapy, vol. 10, no. 3, pp. 365–386, 2020

2020
[10]

Machine learning and health care disparities in dermatology,

Adewole S Adamson and Avery Smith, “Machine learning and health care disparities in dermatology,”JAMA dermatology, vol. 154, no. 11, pp. 1247–1248, 2018

2018
[11]

An analysis of data variation and bias in image-based dermatological datasets for machine learning clas- sification,

Emanoel Santos et al., “An analysis of data variation and bias in image-based dermatological datasets for machine learning clas- sification,”arXiv preprint arXiv:2501.08962, 2025

2025 arXiv
[12]

Continual adaptation of visual representations via domain randomization and meta-learning,

Riccardo V olpi, Diane Larlus, and Gr ´egory Rogez, “Continual adaptation of visual representations via domain randomization and meta-learning,” inProceedings of the IEEE/CVF Confer- ence on Computer Vision and Pattern Recognition, 2021, pp. 4443–4453

2021
[13]

PAD-UFES-20: A skin lesion dataset composed of patient data and clinical images collected from smartphones,

Andre G C Pacheco et al., “PAD-UFES-20: A skin lesion dataset composed of patient data and clinical images collected from smartphones,”Data Brief, vol. 32, no. 106221, pp. 106221, Oct. 2020

2020
[14]

Disparities in dermatology AI perfor- mance on a diverse, curated clinical image set,

Roxana Daneshjou et al., “Disparities in dermatology AI perfor- mance on a diverse, curated clinical image set,”Sci. Adv., vol. 8, no. 32, pp. eabq6147, Aug. 2022

2022
[15]

Al- bumentations: fast and flexible image augmentations,

Alexander Buslaev, Vladimir I Iglovikov, Eugene Khvedchenya, Alex Parinov, Mikhail Druzhinin, and Alexandr A Kalinin, “Al- bumentations: fast and flexible image augmentations,”Informa- tion, vol. 11, no. 2, pp. 125, 2020

2020
[16]

The ham10000 dataset, a large collection of multi-source dermato- scopic images of common pigmented skin lesions,

Philipp Tschandl, Cliff Rosendahl, and Harald Kittler, “The ham10000 dataset, a large collection of multi-source dermato- scopic images of common pigmented skin lesions,”Scientific data, vol. 5, no. 1, pp. 1–9, 2018

2018

Pith tools

Reviewed August 2, 2026 · model on record in the stance chip above.

[1] [1]

This gap remains largely from biases in image acquisition and dataset composition, which shape the visual features learned during training and can lead to unreliable predictions

INTRODUCTION Deep learning has shown strong potential for supporting skin cancer diagnosis [1], yet current models often fail to generalize in practical clinical settings [2]. This gap remains largely from biases in image acquisition and dataset composition, which shape the visual features learned during training and can lead to unreliable predictions. Al...

[2] [2]

Ideally, f(x)should emulate the human decision,H(x)→y

PROBLEM FORMULA TION In supervised learning, we aim to approximate a functionf(x)that predicts a labelˆyfrom a labeled setS={(x i, yi)}N i=1. Ideally, f(x)should emulate the human decision,H(x)→y. However, since the true data distributionp(x, y)is unknown, the sample setS may contain biases or noise, which limits how closelyfcan approxi- mateH. We express...

[3] [3]

Motivation Despite advances in vision algorithms, adoption in dermatological practice remains limited due to variability in imaging conditions

METHODOLOGY 3.1. Motivation Despite advances in vision algorithms, adoption in dermatological practice remains limited due to variability in imaging conditions. These challenges arise from task-related noise inherent to clinical workflows. In this study, we analyze how these factors affect model decisions and introduce a domain-adaptation pipeline designe...

arXiv 2026

[4] [4]

These models were optimized to classify in dermoscopic images

EXPERIMENTS AND DISCUSSIONS We trained the model on the HAM10000 [8], a large collection captured with magnifying devices that highlight fine lesion details. These models were optimized to classify in dermoscopic images. However, our goal is to apply them in clinical scenarios, using datasets such as PAD-UFES-20 [5] and DDI [6]. These datasets are compara...

[5] [5]

Traditional machine learning approaches often struggle with domain variations, which critically impact real-world reliabil- ity

CONCLUSION In this work, we introduce a training strategy to mitigate perfor- mance degradation in clinical models for dermatological lesion as- sessment. Traditional machine learning approaches often struggle with domain variations, which critically impact real-world reliabil- ity. To address this, we propose a contrastive pre-training strat- egy that en...

[6] [6]

Ethical approval was not required, as confirmed by the license attached to the open- access data

COMPLIANCE WITH ETHICAL STANDARDS This research study was conducted retrospectively using human sub- ject data made available in open access [8, 5, 6]. Ethical approval was not required, as confirmed by the license attached to the open- access data

[7] [7]

ACKNOWLEDGMENTS This work was partially supported by INES.IA (National Institute of Science and Technology for Software Engineering Based on and for Artificial Intelligence) www.ines.org.br, CNPq grant 408817/2024-

2024

[8] [8]

8,248, Table 1

The project was supported by the Ministry of Science, Technol- ogy, and Innovation of Brazil, with resources from Law No. 8,248, Table 1. Comparison of naive training, traditional fine-tuning with augmentation, and the proposed contrastive pre-training (CT-pretrain) and guided-tuning (GT) strategies. We replicate the baseline methods reported in [5, 6] an...

arXiv 1991