VB-Mitigator: An Open-source Framework for Evaluating and Advancing Visual Bias Mitigation

Christos Diou; Christos Koutlis; Ioannis Sarridis; Symeon Papadopoulos

REVIEW 4 major objections 6 minor 34 references

VB-Mitigator: An Open-source Framework for Evaluating and Advancing Visual Bias Mitigation

T0 review · 4 major / 6 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read VB-Mitigator unifies 12 visual bias mitigation methods and 7 datasets under one evaluation protocol.

desk verdict A genuinely useful integration effort and benchmark, but the paper's central comparative claims rest on an unverifiable code snapshot and tuning fairness that the manuscript does not yet back up. read the letter →

arxiv 2507.18348 v1 pith:3DUAP3HT submitted 2025-07-24 cs.CV

Ioannis Sarridis , Christos Koutlis , Symeon Papadopoulos , Christos Diou This is my paper · ORCID

classification cs.CV

keywords visualbiasmitigationfairnessincomputervisionspuriouscorrelationsbenchmarkframeworkworst-groupaccuracybias-conflictingdebiasingmethods

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

This paper introduces VB-Mitigator, an open-source software framework that brings twelve established visual bias mitigation methods and seven benchmark datasets under one roof. The aim is to end the field's fragmentation, where every paper uses its own implementation, datasets, and metrics, making fair comparisons nearly impossible. If the framework works as claimed, researchers can develop a new mitigation method by re-implementing only the pipeline stages their method touches, and evaluate it against twelve baselines under a single protocol. The paper backs the framework with a first standardized comparison, reporting worst-group and average accuracy for most datasets and accuracy across seven ImageNet9 test variants for bias-label-unaware methods.

What carries the argument

The architectural core is a set of abstract interfaces grouped around a central BaseTrainer class, which defines every stage of the training pipeline: dataset handling, model setup, criterion, optimizer, scheduler, metric computation, logging, and checkpointing. New mitigation methods inherit from BaseTrainer and re-implement only the pipeline components where their intervention happens, such as custom loss terms, auxiliary bias-capturing models, or dataloader reweighting. Dataset builder modules return metadata dictionaries (number of classes, protected attributes, subgroups) so models, metrics, and training can be configured automatically. This abstraction is what allows twelve methodologically diverse approaches to run under one codebase and one evaluation protocol.

What would settle it

Re-run any single method, say MAVias or GroupDRO, using the authors' original released code and hyperparameters, and compare the resulting worst-group accuracy on Waterbirds and UrbanCars to the numbers in Tables 3 and 4; a large discrepancy would indicate that the framework's reimplementation or tuning, rather than the method, drives the reported ranking.

Watch

Extended reading notes

Core claim

VB-Mitigator claims to provide a unified, extensible research environment for visual bias mitigation: 12 established methods (GroupDRO, DI, EnD, BB, BAdd, LfF, SD, JTT, SoftCon, Debian, FLAC/FLAC-B, and MAVias), 7 datasets spanning synthetic, demographic, background, multi-attribute, and unknown-bias settings, and metrics tailored to fairness evaluation. Using one evaluation protocol across these resources, the paper produces comparative tables in which bias-label-aware methods such as DI, BAdd, and BB generally achieve the highest worst-group accuracy on datasets with known biases, while MAVias, SD, and JTT show the strongest generalization on ImageNet9, where biases are unknown. The paper also recommends worst-group accuracy and average accuracy as the primary metrics for explicitly biased datasets, arguing that accuracy alone and bias-conflict accuracy fail to capture multi-attribute or subgroup disparities.

Load-bearing premise

The comparison's validity rests on the assumption that all twelve methods are faithfully reimplemented from their original papers and tuned with equally appropriate per-dataset hyperparameters, so the observed differences in the tables reflect the methods themselves rather than implementation choices.

Editorial extensions

If this is right

New mitigation methods can be plugged into VB-Mitigator by implementing only the pipeline stages they modify, making the 12 existing methods an immediate baseline set.
Published results from different studies become comparable when they use the framework's protocol, because datasets, metrics, model architectures, and hyperparameters are fixed.
On datasets with known biases, worst-group accuracy and average accuracy are the metrics to report; the results show that bias-label-aware methods generally lead on these metrics.
On datasets with unknown biases, where only bias-label-unaware methods apply, the ImageNet9 results provide reference points for background- and texture-invariance across seven test-set transformations.
The framework's determinism settings and checkpointing support reproducibility of future experiments, within hardware-driven numerical variation.

Reading between the lines

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

If VB-Mitigator becomes a common testbed, its Tables 3 and 4 are likely to be cited as the default comparisons, which raises the stakes of implementation fidelity: a single method that was tuned more or less carefully than the others could skew the field's choice of debiasing technique.
The framework's design points toward a natural next step the paper mentions only as future work: using foundation models to discover bias attributes in general-purpose datasets, which would allow worst-group evaluation beyond the seven current benchmarks.
The instability reported for SoftCon may be a signal about the method's sensitivity to its auxiliary model rather than a definitive ranking, and the framework's configuration system makes that hypothesis directly testable by sweeping its loss weight.

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.

Desk Editor's Note

read the letter

The useful core here is real: a single PyTorch codebase that wraps 12 mitigation methods and 7 datasets behind one trainer/metric/config interface, plus a first pass at a unified evaluation protocol (WGA/AvgAcc, ImageNet9's seven test sets). That is exactly the kind of infrastructure the field needs, and the modularity description is credible. The five-seed runs with standard deviations are also a step up from much of this literature. Credit where due: this is engineering work, and it looks like solid engineering.

The soft spots are in the empirical claims, and they are proportionate to how much the paper leans on them. Tables 3 and 4 are the evidence that the framework gives fair comparisons, but the paper gives no external anchor: no comparison between VB-Mitigator's reproductions and the numbers in the original method papers, no commit hash or code snapshot, and no per-method notes about which details had to be adapted. The stress-test note is right to flag the self-evaluation issue: BAdd, FLAC/FLAC-B, and MAVias are the authors' own methods, their per-dataset hyperparameters are listed in detail, and MAVias happens to top several columns. That does not prove foul play, but it does mean the benchmark's central claim of fair, standardized comparison is currently taken on faith. Also, Table 4 has no vanilla baseline, which makes it hard to tell how much each BLU method actually helps on ImageNet9.

I do not think the paper is incoherent, and the missing baseline and missing code hash are fixable rather than fatal. The framework itself is valuable regardless of whether the benchmark rankings shift by a few points after independent reimplementation. But as submitted, the comparative conclusions are conditional on implementation fidelity and tuning fairness that the manuscript does not yet demonstrate. The paper would benefit from a short reproduction table against original results and a pinned release.

Who is this for? Anyone working on visual bias mitigation who wants a starting codebase, and anyone looking for a quick landscape of how BLU/BLA methods behave across the standard datasets. A serious referee should see it; the architecture and benchmark are useful even if the rankings need scrutiny. My recommendation: send it out, and make the review focus on the code snapshot, the vanilla baseline, and the reproduction evidence. I would not cite the benchmark numbers for a method comparison until those are in place, but I would point people to the framework.

Referee Report

4 major / 6 minor

Summary. The paper introduces VB-Mitigator, an open-source PyTorch framework intended to standardize the development, evaluation, and comparison of visual bias mitigation methods. It describes a modular architecture with components for datasets, mitigators, models, metrics, tools, and configuration, and reports that the framework currently integrates 12 established mitigation methods (5 bias-label-aware and 7 bias-label-unaware) and 7 datasets. The empirical part presents comparative results on Biased-CelebA, Waterbirds, UrbanCars, and ImageNet9, using worst-group accuracy and average accuracy for the first three and accuracy across seven ImageNet9 test variations for the last. The paper also makes recommendations for evaluation practices and discusses limitations and ethical considerations.

Significance. If the framework is indeed made available with a reproducible snapshot, VB-Mitigator could become a useful community resource for reducing fragmentation in visual bias mitigation research. The manuscript's explicit reporting of hyperparameters and its use of five seeds is a positive step, and the BLA/BLU taxonomy and the architecture description are clearly presented. However, the central empirical contribution is not yet independently verifiable from the manuscript: there is no code snapshot or commit hash, no vanilla baseline is reported, and the comparison includes several methods from the same authors without evidence of a common tuning protocol. The claimed "fair, unified comparison" is therefore conditionally supported; the framework itself is promising, but the benchmark conclusions need additional support.

major comments (4)

[§1, §6.2] The manuscript provides only a GitHub URL and no commit hash, DOI, or code snapshot. Because the central contribution is a reproducible evaluation framework, Tables 3 and 4 cannot be checked or extended without a stable version of the code. Please provide a permanent snapshot (e.g., Zenodo) with the exact commit hash and a reproducibility statement documenting package versions and environment details.
[Tables 3 and 4] Neither table includes a vanilla (no-mitigation) baseline. Without a standard empirical risk minimization row, the reader cannot determine whether the compared methods improve on ordinary training, and the relative ranking of methods cannot be interpreted as mitigation gains. Please add the vanilla baseline under the same protocol to both tables; for ImageNet9, this would also clarify whether methods like SD and MAVias actually reduce background dependence relative to a standard ResNet50.
[§6.2] The paper compares several methods that are the authors' own (BAdd, FLAC/FLAC-B, MAVias) and reports per-dataset hyperparameters for FLAC and MAVias, while no per-dataset hyperparameters are listed for LfF or DI and no tuning budgets are given for any method. If the authors' methods received more careful per-dataset tuning, the consistent advantage of methods such as MAVias (e.g., Waterbirds WG Acc 95.90 vs. DI 91.64 in Table 3) could reflect tuning effort rather than method quality. Please report the hyperparameter search procedure or validation-based selection rule for every method, and include a comparison of VB-Mitigator's reproduced numbers with the original publications to demonstrate implementation fidelity.
[§6.1, §6.2] The paper does not state how checkpoints are selected for the reported numbers. If the checkpoint is chosen using the test set or the reported evaluation metric, the results are over-optimistic. Please specify the validation split and model selection rule used for each dataset (or state that a fixed epoch schedule is used), and describe how the primary evaluation metric for checkpoint selection is set for each method.

minor comments (6)

[§3.1] The section heading contains a typo: "Pleliminary" should be "Preliminary".
[§3.2] In the LfF description, "the weight assigned sample i" should read "the weight assigned to sample i".
[§5] The BCA definition contains a subject-verb agreement error and a duplicated phrase: "attempt to focus on the underre-presented groups in the data the data" should be corrected to "attempts to focus on the under-represented groups in the data".
[§3.2, §6.2] The term for MAVias is inconsistently written as "foundational models" and "foundation models"; please unify the terminology.
[Table 4] The arrow notation (↑/↓) in Table 4 is not explained in the text or caption; please add a note that for NO-FG, ONLY-BG-B, and ONLY-BG-T lower accuracy is better, while for the other variants higher accuracy is better.
[References] Some references are incomplete, e.g., [1] lists only the first author; please provide full author lists for all references.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: VB-Mitigator is a framework and empirical benchmark paper with no derivation that reduces to its own inputs.

full rationale

VB-Mitigator is an engineering and benchmarking paper rather than a derivation paper. The central deliverable is an open-source framework, and the supporting evidence consists of empirical comparison tables produced by running published mitigation methods under a stated protocol. Section 3's method descriptions are summaries of prior published algorithms, and Tables 3 and 4 report measured worst-group/accuracy numbers under the evaluation protocol of Sections 6.1 and 6.2; no output quantity is defined in terms of an input quantity through a fitted parameter, and no uniqueness theorem or prior result is invoked to force a conclusion. The authors' own methods (BAdd, FLAC, MAVias) are cited as prior published work [22,27,28]; these self-citations are normal and not load-bearing because the framework's existence, modularity, and extensibility claims do not depend on those methods being superior. The main caveat is implementation fidelity and tuning fairness: Section 6.2 asserts that method-specific hyperparameters were configured following the values recommended in the original publications, but the paper does not include a side-by-side reproduction table against the original papers' reported numbers. That is a validity risk for the benchmark, not a circularity, because the comparison results are not equal to the inputs by construction. The limitations section openly notes that future methods may pose integration challenges and that current methods do not guarantee fair models, which further indicates a non-circular, engineering-oriented contribution.

Assumptions & free parameters 7 free parameters · 3 assumptions · 0 invented entities

The paper introduces no new entities or formal derivations. Its central benchmark depends on many method-specific hyperparameters carried from prior papers, and on the assumption that bias proxies and fairness metrics used by the framework capture the biases they claim to measure.

free parameters (7)

MAVias lambda_1 and lambda_2 = (0.01,0.5) CelebA, (0.05,0.6) Waterbirds, (0.01,0.4) UrbanCars, (0.001,0.7) ImageNet9
Set per dataset in Section 6.2 following the original MAVias paper; directly affect the MAVias results in Tables 3 and 4.
FLAC lambda = 30,000 / 10,000 / 10,000 / 100 for CelebA / Waterbirds / UrbanCars / ImageNet9
Set per dataset in Section 6.2; FLAC-B results depend on these choices.
JTT reweighting and optimizer hyperparameters = upweight 100, learning rate 1e-5, weight decay 1
Taken from the JTT paper; govern which samples are treated as bias-conflicting.
GroupDRO robust step size = 0.01
Set in Section 6.2; controls worst-group optimization.
SD regularization lambda = 0.1
Set in Section 6.2; controls spectral decoupling regularization strength.
EnD lambda_1 and lambda_2 = 1 and 1
Set in Section 6.2; balances disentanglement and entanglement losses.
SoftCon cross-entropy weight = 0.01
Set in Section 6.2; balances constrastive and classification losses.

assumptions (3)

domain assumption Spurious correlations are adequately captured by explicit bias labels or by bias-capturing auxiliary models, and worst-group accuracy and average accuracy are valid primary measures of fairness.
The framework's evaluation protocol in Sections 5 and 6.1 treats worst-group and average accuracy as the main fairness metrics, which presumes subgroup annotations are correct and group-wise accuracy captures the intended notion of fairness.
domain assumption Bias-capturing models used by bias-label-unaware methods, such as LfF, Debian, SoftCon, FLAC-B, and MAVias, reliably encode the bias attribute without access to bias labels.
The method descriptions in Section 3.2 rely on auxiliary models or foundation-model tags to infer biases; if those signals are weak, the reported mitigation results would not generalize.
domain assumption Fixed random seeds and deterministic CUDA settings are sufficient to reproduce published numbers across hardware.
Section 2.2 claims reproducibility through seeding and deterministic mode, but also admits that hardware and CUDA version variation can cause discrepancies.

how reviews work

0 comments

Cite this review

Pith. "Pith review of VB-Mitigator: An Open-source Framework for Evaluating and Advancing Visual Bias Mitigation." pith.science (2026). https://pith.science/paper/3DUAP3HT

@misc{pith2026250718348,
  author       = {Pith},
  title        = {Pith review of: VB-Mitigator: An Open-source Framework for Evaluating and Advancing Visual Bias Mitigation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/3DUAP3HT}},
  note         = {Machine review of arXiv:2507.18348}
}

read the original abstract

Bias in computer vision models remains a significant challenge, often resulting in unfair, unreliable, and non-generalizable AI systems. Although research into bias mitigation has intensified, progress continues to be hindered by fragmented implementations and inconsistent evaluation practices. Disparate datasets and metrics used across studies complicate reproducibility, making it difficult to fairly assess and compare the effectiveness of various approaches. To overcome these limitations, we introduce the Visual Bias Mitigator (VB-Mitigator), an open-source framework designed to streamline the development, evaluation, and comparative analysis of visual bias mitigation techniques. VB-Mitigator offers a unified research environment encompassing 12 established mitigation methods, 7 diverse benchmark datasets. A key strength of VB-Mitigator is its extensibility, allowing for seamless integration of additional methods, datasets, metrics, and models. VB-Mitigator aims to accelerate research toward fairness-aware computer vision models by serving as a foundational codebase for the research community to develop and assess their approaches. To this end, we also recommend best evaluation practices and provide a comprehensive performance comparison among state-of-the-art methodologies.

Figures

Figures reproduced from arXiv: 2507.18348 by the authors.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

34 extracted references · 24 canonical work pages

[1]

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

Alexey Dosovitskiy. An image is worth 16x16 words: Transformers for image recognition at scale. arXiv preprint arXiv:2010.11929, 2020

arXiv 2010
[2]

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. In International conference on machine learning , pages 1597–1607. PmLR, 2020

work page 2020
[3]

Generative adversarial networks in computer vision: A survey and taxonomy

Zhengwei Wang, Qi She, and Tomas E Ward. Generative adversarial networks in computer vision: A survey and taxonomy. ACM Computing Surveys (CSUR), 54(2):1–38, 2021

work page 2021
[4]

A survey on bias and fairness in machine learning

Ninareh Mehrabi, Fred Morstatter, Nripsuta Saxena, Kristina Lerman, and Aram Galstyan. A survey on bias and fairness in machine learning. ACM computing surveys (CSUR), 54(6):1–35, 2021

2021
[5]

A survey on bias in visual datasets

Simone Fabbrizzi, Symeon Papadopoulos, Eirini Ntoutsi, and Ioannis Kompatsiaris. A survey on bias in visual datasets. Computer Vision and Image Understanding, 223:103552, 2022

work page 2022
[6]

Frcsyn challenge at cvpr 2024: Face recognition challenge in the era of synthetic data

Ivan DeAndres-Tame, Ruben Tolosana, Pietro Melzi, Ruben Vera-Rodriguez, Minchul Kim, Christian Rathgeb, Xiaoming Liu, Aythami Morales, Julian Fierrez, Javier Ortega-Garcia, et al. Frcsyn challenge at cvpr 2024: Face recognition challenge in the era of synthetic data. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pag...

work page 2024
[7]

Bias in data-driven artificial intelligence systems—an introductory survey

Eirini Ntoutsi, Pavlos Fafalios, Ujwal Gadiraju, Vasileios Iosifidis, Wolfgang Nejdl, Maria-Esther Vidal, Salvatore Ruggieri, Franco Turini, Symeon Papadopoulos, Emmanouil Krasanakis, et al. Bias in data-driven artificial intelligence systems—an introductory survey. Wiley Interdisciplinary Reviews: Data Mining and Knowledge Discovery, 10(3):e1356, 2020

work page 2020
[8]

Facex: Understanding face attribute classifiers through summary model explanations

Ioannis Sarridis, Christos Koutlis, Symeon Papadopoulos, and Christos Diou. Facex: Understanding face attribute classifiers through summary model explanations. In Proceedings of the 2024 International Conference on Multimedia Retrieval, pages 758–766, 2024

work page 2024

Show all 34 references

[9]

Spurious correlations in machine learning: A survey

Wenqian Ye, Guangtao Zheng, Xu Cao, Yunsheng Ma, and Aidong Zhang. Spurious correlations in machine learning: A survey. arXiv preprint arXiv:2402.12715, 2024

2024
[10]

Unbiased supervised contrastive learning

Carlo Alberto Barbano, Benoit Dufumier, Enzo Tartaglione, Marco Grangetto, and Pietro Gori. Unbiased supervised contrastive learning. arXiv preprint arXiv:2211.05568, 2022

2022 arXiv
[11]

A whac-a-mole dilemma: Shortcuts come in multiples where mitigating one amplifies others

Zhiheng Li, Ivan Evtimov, Albert Gordo, Caner Hazirbas, Tal Hassner, Cristian Canton Ferrer, Chenliang Xu, and Mark Ibrahim. A whac-a-mole dilemma: Shortcuts come in multiples where mitigating one amplifies others. In Proceedings of the IEEE/CVF Conference on Computer Vision a...

2023
[12]

Frcsyn-ongoing: Benchmarking and comprehensive evaluation of real and synthetic data to improve face recognition systems

Pietro Melzi, Ruben Tolosana, Ruben Vera-Rodriguez, Minchul Kim, Christian Rathgeb, Xiaoming Liu, Ivan DeAndres-Tame, Aythami Morales, Julian Fierrez, Javier Ortega-Garcia, et al. Frcsyn-ongoing: Benchmarking and comprehensive evaluation of real and synthetic data to improve f...

2024
[13]

Towards fair face verification: An in-depth analysis of demographic biases

Ioannis Sarridis, Christos Koutlis, Symeon Papadopoulos, and Christos Diou. Towards fair face verification: An in-depth analysis of demographic biases. In Proceedings of the International Workshops of ECML PKDD , 2023

2023
[14]

Learning de-biased representa- tions with biased representations

Hyojin Bahng, Sanghyuk Chun, Sangdoo Yun, Jaegul Choo, and Seong Joon Oh. Learning de-biased representa- tions with biased representations. In International Conference on Machine Learning , pages 528–539. PMLR, 2020. 10

2020
[15]

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

2016
[16]

Efficientnet: Rethinking model scaling for convolutional neural networks

Mingxing Tan and Quoc Le. Efficientnet: Rethinking model scaling for convolutional neural networks. In International conference on machine learning , pages 6105–6114. PMLR, 2019

2019
[17]

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. InProceedings of the IEEE/CVF international conference on computer vision, pages 10012–10022, 2021

2021
[18]

Distributionally robust neural networks for group shifts: On the importance of regularization for worst-case generalization

Shiori Sagawa, Pang Wei Koh, Tatsunori B Hashimoto, and Percy Liang. Distributionally robust neural networks for group shifts: On the importance of regularization for worst-case generalization. arXiv preprint arXiv:1911.08731, 2019

1911 arXiv
[19]

Towards fairness in visual recognition: Effective strategies for bias mitigation

Zeyu Wang, Klint Qinami, Ioannis Christos Karakozis, Kyle Genova, Prem Nair, Kenji Hata, and Olga Rus- sakovsky. Towards fairness in visual recognition: Effective strategies for bias mitigation. In Proceedings of the IEEE/CVF conference on computer vision and pattern recogniti...

2020
[20]

End: Entangling and disentangling deep representations for bias correction

Enzo Tartaglione, Carlo Alberto Barbano, and Marco Grangetto. End: Entangling and disentangling deep representations for bias correction. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 13508–13517, 2021

2021
[21]

Unbiased classification through bias-contrastive and bias-balanced learning

Youngkyu Hong and Eunho Yang. Unbiased classification through bias-contrastive and bias-balanced learning. Advances in Neural Information Processing Systems , 34:26449–26461, 2021

2021
[22]

Badd: Bias mitigation through bias addition

Ioannis Sarridis, Christos Koutlis, Symeon Papadopoulos, and Christos Diou. Badd: Bias mitigation through bias addition. arXiv preprint arXiv:2408.11439, 2024

2024 arXiv
[23]

Learning from failure: De-biasing classifier from biased classifier

Junhyun Nam, Hyuntak Cha, Sungsoo Ahn, Jaeho Lee, and Jinwoo Shin. Learning from failure: De-biasing classifier from biased classifier. Advances in Neural Information Processing Systems , 33:20673–20684, 2020

2020
[24]

Gradient starvation: A learning proclivity in neural networks

Mohammad Pezeshki, Oumar Kaba, Yoshua Bengio, Aaron C Courville, Doina Precup, and Guillaume Lajoie. Gradient starvation: A learning proclivity in neural networks. Advances in Neural Information Processing Systems, 34:1256–1272, 2021

2021
[25]

Just train twice: Improving group robustness without training group information

Evan Z Liu, Behzad Haghgoo, Annie S Chen, Aditi Raghunathan, Pang Wei Koh, Shiori Sagawa, Percy Liang, and Chelsea Finn. Just train twice: Improving group robustness without training group information. In International Conference on Machine Learning, pages 6781–6792. PMLR, 2021

2021
[26]

Discover and mitigate unknown biases with debiasing alternate networks

Zhiheng Li, Anthony Hoogs, and Chenliang Xu. Discover and mitigate unknown biases with debiasing alternate networks. In European Conference on Computer Vision, pages 270–288. Springer, 2022

2022
[27]

Flac: Fairness-aware representation learning by suppressing attribute-class associations

Ioannis Sarridis, Christos Koutlis, Symeon Papadopoulos, and Christos Diou. Flac: Fairness-aware representation learning by suppressing attribute-class associations. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2024

2024
[28]

Mavias: Mitigate any visual bias

Ioannis Sarridis, Christos Koutlis, Symeon Papadopoulos, and Christos Diou. Mavias: Mitigate any visual bias. arXiv preprint arXiv:2412.06632, 2024

2024
[29]

Supervised contrastive learning

Prannay Khosla, Piotr Teterwak, Chen Wang, Aaron Sarna, Yonglong Tian, Phillip Isola, Aaron Maschinot, Ce Liu, and Dilip Krishnan. Supervised contrastive learning. Advances in neural information processing systems , 33:18661–18673, 2020

2020
[30]

Age progression/regression by conditional adversarial autoencoder

Zhang Zhifei, Song Yang, and Qi Hairong. Age progression/regression by conditional adversarial autoencoder. In IEEE Conference on Computer Vision and Pattern Recognition (CVPR) . IEEE, 2017

2017
[31]

Deep learning face attributes in the wild

Ziwei Liu, Ping Luo, Xiaogang Wang, and Xiaoou Tang. Deep learning face attributes in the wild. In Proceedings of International Conference on Computer Vision (ICCV) , December 2015

2015
[32]

Noise or signal: The role of image backgrounds in object recognition

Kai Yuanqing Xiao, Logan Engstrom, Andrew Ilyas, and Aleksander Madry. Noise or signal: The role of image backgrounds in object recognition. In International Conference on Learning Representations , 2021

2021
[33]

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
[34]

Say my name: a model’s bias discovery framework

Massimiliano Ciranni, Luca Molinaro, Carlo Alberto Barbano, Attilio Fiandrotti, Vittorio Murino, Vito Paolo Pas- tore, and Enzo Tartaglione. Say my name: a model’s bias discovery framework. arXiv preprint arXiv:2408.09570, 2024. 11

2024

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.

[1] [1]

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

Alexey Dosovitskiy. An image is worth 16x16 words: Transformers for image recognition at scale. arXiv preprint arXiv:2010.11929, 2020

arXiv 2010

[2] [2]

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. In International conference on machine learning , pages 1597–1607. PmLR, 2020

work page 2020

[3] [3]

Generative adversarial networks in computer vision: A survey and taxonomy

Zhengwei Wang, Qi She, and Tomas E Ward. Generative adversarial networks in computer vision: A survey and taxonomy. ACM Computing Surveys (CSUR), 54(2):1–38, 2021

work page 2021

[4] [4]

A survey on bias and fairness in machine learning

Ninareh Mehrabi, Fred Morstatter, Nripsuta Saxena, Kristina Lerman, and Aram Galstyan. A survey on bias and fairness in machine learning. ACM computing surveys (CSUR), 54(6):1–35, 2021

2021

[5] [5]

A survey on bias in visual datasets

Simone Fabbrizzi, Symeon Papadopoulos, Eirini Ntoutsi, and Ioannis Kompatsiaris. A survey on bias in visual datasets. Computer Vision and Image Understanding, 223:103552, 2022

work page 2022

[6] [6]

Frcsyn challenge at cvpr 2024: Face recognition challenge in the era of synthetic data

Ivan DeAndres-Tame, Ruben Tolosana, Pietro Melzi, Ruben Vera-Rodriguez, Minchul Kim, Christian Rathgeb, Xiaoming Liu, Aythami Morales, Julian Fierrez, Javier Ortega-Garcia, et al. Frcsyn challenge at cvpr 2024: Face recognition challenge in the era of synthetic data. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pag...

work page 2024

[7] [7]

Bias in data-driven artificial intelligence systems—an introductory survey

Eirini Ntoutsi, Pavlos Fafalios, Ujwal Gadiraju, Vasileios Iosifidis, Wolfgang Nejdl, Maria-Esther Vidal, Salvatore Ruggieri, Franco Turini, Symeon Papadopoulos, Emmanouil Krasanakis, et al. Bias in data-driven artificial intelligence systems—an introductory survey. Wiley Interdisciplinary Reviews: Data Mining and Knowledge Discovery, 10(3):e1356, 2020

work page 2020

[8] [8]

Facex: Understanding face attribute classifiers through summary model explanations

Ioannis Sarridis, Christos Koutlis, Symeon Papadopoulos, and Christos Diou. Facex: Understanding face attribute classifiers through summary model explanations. In Proceedings of the 2024 International Conference on Multimedia Retrieval, pages 758–766, 2024

work page 2024