AWARE-NET: Adaptive Weighted Averaging for Robust Ensemble Network in Deepfake Detection

Iqra Tariq; Mishal Zulfiqar; Muhammad Salman; Muqadas Jalal; Sami Aujla; Sumbal Fatima

REVIEW 3 major objections 4 minor 39 references

AWARE-NET: Adaptive Weighted Averaging for Robust Ensemble Network in Deepfake Detection

T0 review · 3 major / 4 minor · reviewed 2026-08-16 · deepseek-v4-flash

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

Pith's one-line read AWARE-NET claims a two-tier ensemble with learnable weights lifts cross-dataset deepfake detection from 30.31% to 88.20% AUC.

desk verdict The cross-dataset AUC of 88.20% is arithmetically impossible given Table 3 and Eq. 7; the paper's only plausible result is a small intra-dataset gain near saturation. read the letter →

arxiv 2505.00312 v1 pith:27GQUNRA submitted 2025-05-01 cs.CV

Muhammad Salman , Iqra Tariq , Mishal Zulfiqar , Muqadas Jalal , Sami Aujla , Sumbal Fatima This is my paper · ORCID

classification cs.CV

keywords deepfakedetectionensemblelearninglearnableweightscross-datasetgeneralizationFaceForensics++CelebDF-v2two-tierfusionXception

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 introduces AWARE-NET, a two-tier ensemble for deepfake detection that combines three instances each of Xception, Res2Net101, and EfficientNet-B7: the first tier averages predictions within each architecture family, and the second tier learns a softmax-weighted combination of the three family outputs through backpropagation. The authors report near-perfect intra-dataset results (99.22% AUC on FaceForensics++, 100% on CelebDF-v2) and, as the central finding, a cross-dataset AUC of 88.20% when models trained on FaceForensics++ are tested on CelebDF-v2, where the best individual architecture reaches only 30.31%. If correct, this shows that a learnably weighted ensemble can convert individually failing detectors into a strong generalizing one. The framework is fully differentiable and yields interpretable per-architecture weights.

What carries the argument

The central object is the two-tier fusion rule. First-tier: for each architecture family $A \in \{\text{Xception}, \text{Res2Net101}, \text{EfficientNet-B7}\}$, three independently initialized instances produce sigmoid outputs $\sigma(M_i^A(x))$ whose mean is $p_A(x)$ (Eq. 5). Second-tier: a learnable parameter vector $w=(w_1,w_2,w_3)$ is passed through softmax to give $\alpha=\mathrm{softmax}(w)$, and the final prediction is $y(x)=\sum_i \alpha_i p_i(x)$ (Eq. 7). The machinery is fully differentiable: the gradient of the loss with respect to $w_i$ (Eq. 8) flows through $\partial y/\partial \alpha_i = p_i(x)$ and the softmax Jacobian, so the ensemble discovers which architecture to trust during training.

What would settle it

Recompute the ensemble's score on the CelebDF-v2 test set as the weighted average of the three architecture-family scores using the learned weights, and verify that the resulting AUC is approximately 88.20% while each family's individual AUC is below 31%. If no positive weighting of those scores yields an AUC above 50%, the reported jump is an artifact of inconsistent evaluation rather than a property of the ensemble.

Watch

Extended reading notes

Core claim

The paper's central claim is that a two-tier ensemble can turn individually poor cross-dataset detectors into a strong one. Trained on FaceForensics++ and tested on CelebDF-v2, AWARE-NET reports an AUC of 88.20% and an F1 score of 93.16%, while every individual architecture — Xception, Res2Net101, and EfficientNet-B7 — achieves an AUC below 31%, i.e., below chance. The authors attribute this to diversity across the three families and to the adaptive weighting mechanism, which learns to emphasize reliable architectures through backpropagation. On the reverse transfer the advantage is smaller but positive (72.52% AUC versus 71.76% for the best single model), and on intra-dataset tests the ensemble reaches 99.22% AUC on FF++ and 100% on CelebDF-v2.

Load-bearing premise

The claim stands on the assumption that the base-model AUCs and the ensemble AUC in Table 3 come from the same test set with identical label orientation; otherwise the convex combination in Eq. 7 could not turn several below-chance scores into an 88% AUC.

Editorial extensions

If this is right

AWARE-NET reaches essentially ceiling performance on intra-dataset tests: 99.22% AUC on FF++ and 100% on CelebDF-v2 without augmentation, with F1 scores above 98% in both cases.
On the FF++ to CelebDF-v2 transfer, the ensemble reports 88.20% AUC versus 30.31% for the best individual network, implying that the learned combination exploits failure complementarity.
On the reverse transfer (CelebDF-v2 to FF++), the ensemble improves AUC to 72.52% from the best single model's 71.76%, a positive but much smaller gain.
Data augmentation helps intra-dataset performance but slightly hurts cross-dataset transfer (69.66% vs 72.52% AUC in one direction), indicating that source-domain augmentation strategies do not transfer.

Reading between the lines

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

The 57.89-point AUC jump from below-chance individuals to an 88% ensemble is extraordinary for a convex combination with fixed positive weights; a straightforward check is whether the base-model AUCs in Table 3 share the same label orientation as the ensemble row.
If the result reproduces under a consistent evaluation protocol, it would suggest that model diversity alone can act as a form of unsupervised domain adaptation, with the learned weights effectively suppressing unreliable architectures on the source domain.
A natural extension is to record the learned weights $\alpha$ on source data and inspect whether they shift when a small labeled target-domain sample is introduced, turning the frozen ensemble into a few-shot domain-adaptation tool.

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

Bottom line: the headline cross-dataset result is not just surprising, it is inconsistent with the paper's own method. Table 3 reports AWARE-NET at 88.20% AUC on FF++ to CelebDF-v2 while every base model scores below 31% AUC on that same direction. Since Eq. 7 forms the ensemble output as a convex combination (positive softmax weights) of the base sigmoid outputs, the ensemble score is a weighted average of the base scores. If each base model ranks a fake below a real, the ensemble must do the same. The ensemble AUC therefore cannot exceed the sum of the base AUCs, which here is about 79%, and with all bases below chance it would actually be far lower. The only ways to get 88.20% are different label orientation, a different test set, or a different equation than the one reported. The paper does not acknowledge any of these. This kills the central generalization claim.

What the paper does well: the two-tier structure—averaging three instances within each architecture family, then fusing with learned softmax weights—is a reasonable recombination of standard ensemble ideas. The intra-dataset tables look plausible: near-saturated AUCs on FF++ and CelebDF-v2 with small but consistent gains from the ensemble. The method section is straightforwardly written.

Soft spots beyond the fatal one: no code, no error bars, no ablations on the number of instances, and no comparison against the cited ensemble baselines. The augmentation cross-dataset text also mislabels the direction: the 69.66% figure it attributes to FF++ to CelebDF-v2 is actually the reverse transfer, and the actual FF++ to CelebDF-v2 number with augmentation is 22.43%. That kind of sloppiness reinforces the concern that the 88.20% comes from an evaluation inconsistency rather than a real effect. The learned weights are three scalars fit on validation data, which is standard supervised learning, not circular, but also not the dramatic mechanism the paper claims.

Who is this for? Not many people, in its current form. If the authors can release code and produce a single consistent evaluation protocol, the intra-dataset ensemble idea might be a minor contribution. As submitted, the main result is arithmetically impossible, and the paper should not consume referee time until that is resolved.

Referee Report

3 major / 4 minor

Summary. The manuscript proposes AWARE-NET, a two-tier ensemble for deepfake video detection. Three instances each of Xception, Res2Net101, and EfficientNet-B7 are trained; their sigmoid outputs are averaged within each architecture family in the first tier, and a softmax-normalized learnable weight vector combines the three family-level scores in the second tier. Experiments on FaceForensics++ and CelebDF-v2 report near-saturated intra-dataset AUCs of 99.22% and 100.00%, and cross-dataset AUCs of 88.20% and 72.52%. The paper's central claim is that the learned-weight two-tier fusion substantially improves cross-dataset generalization, especially in the FF++ to CelebDF-v2 direction.

Significance. If the cross-dataset numbers were correct, the result would be remarkable, since it would imply that a positive-weight convex combination of three below-chance base detectors (AUCs 24.37%, 24.61%, 30.31%) becomes a highly accurate 88.20% AUC detector. That implication is impossible under the paper's own Eq. (7), as the stress-test upper bound shows. The intra-dataset results are plausible but near saturation and add little over the best base model, so the cross-dataset generalization claim is the main contribution of the paper, and that claim is internally inconsistent. On the positive side, the framework is clearly described, the intra-dataset evaluation uses standard datasets and metrics, and the inconsistency is precisely diagnosable from the provided equations and tables, which gives the authors a concrete path to locate the evaluation error. No code or model weights are released, so the discrepancy cannot currently be resolved by artifact inspection.

major comments (3)

[Section 5.2, Eq. (7), Table 3] The reported cross-dataset AUC of 88.20% for FF++ to CelebDF-v2 is impossible under Eq. (7) given the individual AUCs in Table 3. Eq. (7) defines y(x) = sum_i alpha_i p_i(x) with alpha_i > 0 and sum_i alpha_i = 1. For any fake/real pair, define D_i = p_i(fake) - p_i(real); the ensemble ranks the pair correctly only if sum_i alpha_i D_i > 0. This cannot happen when all three D_i <= 0, so the ensemble can correctly rank a pair only if at least one base model does. Hence AUC_ens <= P(union_i {D_i > 0}) <= sum_i AUC_i = 24.37% + 24.61% + 30.31% = 79.29%, which is strictly less than 88.20%. The reported ensemble AUC therefore requires either a different label orientation, a different test set, or a different fusion formula than Eq. (7) for the AWARE-NET row. This is a load-bearing inconsistency in the central generalization claim.
[Section 5.2, Tables 3 and 4] The augmentation results in Table 4 are mutually inconsistent with Table 3 and with the text. Table 4 reports an AWARE-NET AUC of 22.43% for FF++ to CelebDF-v2 with augmentation, whereas Table 3 reports 88.20% for the same direction without augmentation; this 65.77-point drop is not a plausible augmentation effect. The surrounding text adds further confusion by stating 'FF++ to CelebDF-v2: 69.66% vs. 72.52% without augmentation,' although the 72.52% value in Table 3 appears in the column that the text elsewhere associates with CelebDF-v2 to FF++. The column labels in Tables 3 and 4 must be reconciled with the transfer directions stated in the text before any cross-dataset conclusion can be evaluated.
[Section 6] The concluding paragraph attributes the '57.89% improvement' to the CelebDF-v2 to FF++ transfer. From Table 3, 57.89 = 88.20 - 30.31 belongs to the FF++ to CelebDF-v2 direction, while the CelebDF-v2 to FF++ improvement over the best base model is 72.52 - 71.76 = 0.76%. This mislabeled direction is consistent with the column-orientation problem in Tables 3 and 4 and should be corrected in any revision.

minor comments (4)

[Section 4] The contribution list claims that the framework 'jointly optimizes model parameters and architectural weights,' but the implementation section states that Phase 2 freezes the pre-trained base models and trains only the weight vector w; the wording should be aligned with the actual training procedure.
[Section 5.2] The paper does not state whether AUC and F1 are computed at frame level or video level, nor how many test videos or frames are used in the cross-dataset evaluation; this information is needed to interpret the absolute numbers.
[Section 4] There are typographical errors, such as 'pochs' for 'epochs' in Section 4, and Figure 9's axes and color key are not self-explanatory.
[References] References [29] and [30] appear to duplicate the same survey, and reference [35] is a GitHub repository link rather than a formal citation; the reference formatting should be normalized.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the ensemble prediction is a standard held-out evaluation of a learned weighted average, with no test-set quantities used as inputs.

full rationale

The paper's derivation chain is self-contained in the circularity-relevant sense. The two-tier ensemble is defined by Eqs. (4)-(7): first-tier per-architecture predictions are arithmetic means of three instance sigmoid outputs, and the second-tier output is a softmax-normalized convex combination of those means. The learnable weights w are trained by backpropagation on the source-domain training/validation set (Section 4), not derived from or fitted to the target test set, so the cross-dataset AUCs in Tables 3 and 4 are genuine held-out predictions rather than re-statements of the training objective. No parameter in the method is defined in terms of the reported evaluation metric, no uniqueness theorem or prior result by the same authors is invoked to force the architecture choice, and no known result is relabeled as a new prediction. The reported cross-dataset numbers are surprising and possibly internally inconsistent (e.g., an 88.20% AUC from a convex combination of base models with AUCs of 24-30% is not attainable under the paper's own Eq. (7) if all AUCs share one label orientation and test set), but that inconsistency is a correctness or evaluation-consistency concern, not circularity: the contradiction, if real, would mean the table does not report what Eq. (7) computes, not that the output was used as an input. No circular step is therefore identified.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

The central claim rests on two fitted quantities: the three ensemble weights (learned) and the design choice of three instances per architecture. The most fragile assumption is the consistency and orientation of the evaluation labels across the individual and ensemble rows of Table 3. No new physical or mathematical entities are introduced.

free parameters (3)

learnable ensemble weights w1,w2,w3 = not reported (initialized to [1/3,1/3,1/3])
Tier-2 weights are optimized by backpropagation on training data; the learned values are never reported, and the final performance depends on them.
number of instances per architecture (k=3) = 3
The choice of three independent instances per architecture is made without ablation and directly shapes the ensemble variance reduction claimed in the paper.
frame sampling counts (32 real, 16 fake) = 32 and 16
These counts are chosen by hand in preprocessing and affect the data distribution, but no sensitivity analysis is provided.

assumptions (4)

domain assumption The evaluation protocol applies the same label orientation and metric definitions to all models and datasets.
The paper never states the positive/negative class convention. The contradiction between below-chance individual AUCs and an 88.20% ensemble AUC on the same column suggests this assumption is violated.
domain assumption The train/test split of video frames does not leak frames from the same video across the split.
The paper does not describe whether the split is video-level or frame-level; frame-level splits are a known source of inflated deepfake detection results.
domain assumption The three backbone architectures are used as pretrained feature extractors whose features transfer to face forgery detection.
The paper uses timm models but does not explicitly state ImageNet or other pretraining; the ensemble's performance depends on this transfer.
domain assumption Averaging three instances with different initializations reduces variance without introducing systematic bias.
The paper asserts this as a motivation but does not provide variance measurements or an ablation comparing one instance versus three.

how reviews work

0 comments

Cite this review

Pith. "Pith review of AWARE-NET: Adaptive Weighted Averaging for Robust Ensemble Network in Deepfake Detection." pith.science (2026). https://pith.science/paper/27GQUNRA

@misc{pith2026250500312,
  author       = {Pith},
  title        = {Pith review of: AWARE-NET: Adaptive Weighted Averaging for Robust Ensemble Network in Deepfake Detection},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/27GQUNRA}},
  note         = {Machine review of arXiv:2505.00312}
}

read the original abstract

Deepfake detection has become increasingly important due to the rise of synthetic media, which poses significant risks to digital identity and cyber presence for security and trust. While multiple approaches have improved detection accuracy, challenges remain in achieving consistent performance across diverse datasets and manipulation types. In response, we propose a novel two-tier ensemble framework for deepfake detection based on deep learning that hierarchically combines multiple instances of three state-of-the-art architectures: Xception, Res2Net101, and EfficientNet-B7. Our framework employs a unique approach where each architecture is instantiated three times with different initializations to enhance model diversity, followed by a learnable weighting mechanism that dynamically combines their predictions. Unlike traditional fixed-weight ensembles, our first-tier averages predictions within each architecture family to reduce model variance, while the second tier learns optimal contribution weights through backpropagation, automatically adjusting each architecture's influence based on their detection reliability. Our experiments achieved state-of-the-art intra-dataset performance with AUC scores of 99.22% (FF++) and 100.00% (CelebDF-v2), and F1 scores of 98.06% (FF++) and 99.94% (CelebDF-v2) without augmentation. With augmentation, we achieve AUC scores of 99.47% (FF++) and 100.00% (CelebDF-v2), and F1 scores of 98.43% (FF++) and 99.95% (CelebDF-v2). The framework demonstrates robust cross-dataset generalization, achieving AUC scores of 88.20% and 72.52%, and F1 scores of 93.16% and 80.62% in cross-dataset evaluations.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

39 extracted references · 14 canonical work pages

[1]

Talk to Meta AI on WhatsApp,

“Talk to Meta AI on WhatsApp,” WhatsApp.com. Accessed: Dec. 04,

work page
[2]

Say Hi to My AI

“Say Hi to My AI.” Accessed: Dec. 04, 2024. [Online]. Available: https://newsroom.snap.com/say-hi-to-my-ai

work page 2024
[3]

The Creation and Detection of Deepfakes: A Survey,

Y. Mirsky and W. Lee, “The Creation and Detection of Deepfakes: A Survey,” ACM Comput. Surv., vol. 54, no. 1, p. 7:1 -7:41, Jan. 2021, doi: 10.1145/3425780

doi:10.1145/3425780 2021
[4]

NSA, U.S. Federal Agencies Advise on Deepfake Threats,

“NSA, U.S. Federal Agencies Advise on Deepfake Threats,” National Security Agency/Central Security Service. Accessed: Nov. 30, 2024. [Online]. Available: https://www.nsa.gov/Press -Room/Press- Releases-Statements/Press-Release-View/Article/3523329/nsa-us- federal-agencies-advise-on-deepfake- threats/http%3A%2F%2Fwww.nsa.gov%2FPress-Room%2FPress- Releases-...

arXiv 2024
[5]

Battle of Deep Fakes: Artificial Intelligence Set to Become a Major Threat to the Individual and National Security,

A. Ali, K. F. Khan Ghouri, H. Naseem, T. R. Soomro, W. Mansoor, and A. M. Momani, “Battle of Deep Fakes: Artificial Intelligence Set to Become a Major Threat to the Individual and National Security,” in 2022 International Conference on Cyber Resilience (ICCR) , Oct. 2022, pp. 1–5. doi: 10.1109/ICCR56254.2022.9995821

arXiv 2022
[6]

Deepfakes and the Dangers to National Security and Defence - RSIS

“Deepfakes and the Dangers to National Security and Defence - RSIS.” Accessed: Dec. 04, 2024. [Online]. Available: https://www.rsis.edu.sg/rsis-publication/rsis/deepfakes-and-the- Architectures FF++ CelebDF-v2 AUC Accuracy Precision Recall F1 AUC Accuracy Precision Recall F1 Xception 99.25% 97.04% 97.74% 98.60% 98.17% 100.00% 99.90% 99.91% 99.94% 99.92% R...

work page arXiv 2024
[7]

Protecting World Leaders Against Deep Fakes,

S. Agarwal, H. Farid, Y. Gu, M. He, K. Nagano, and H. Li, “Protecting World Leaders Against Deep Fakes,” in IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Workshops, IEEE, 2019, pp. 38 –45. Accessed: Nov. 30, 2024. [Online]. Available: https://openaccess.thecvf.com/content_CVPRW_2019/html/Media_F orensics/Agarwal_Protecting_World_Le...

work page 2019
[8]

Deepfake in Movies | Facts, Fiction & Future of Film Industry,

teresa_myers, “Deepfake in Movies | Facts, Fiction & Future of Film Industry,” Facia.ai. Accessed: Dec. 04, 2024. [Online]. Available: https://facia.ai/blog/deepfake-in-movies-facts-fiction-future-of-film- industry/

work page 2024

Show all 39 references

[9]

Face/Off: Changing the face of movies with deepfakes,

G. Murphy, D. Ching, J. Twomey, and C. Linehan, “Face/Off: Changing the face of movies with deepfakes,” PLOS ONE, vol. 18, no. 7, p. e0287503, Jul. 2023, doi: 10.1371/journal.pone.0287503

2023 doi
[10]

Imran Khan —Pakistan’s Jailed Ex -Leader—Uses AI Deepfake To Address Online Election Rally,

S. Ray, “Imran Khan —Pakistan’s Jailed Ex -Leader—Uses AI Deepfake To Address Online Election Rally,” Forbes. Accessed: Dec. 04, 2024. [Online]. Available: https://www.forbes.com/sites/siladityaray/2023/12/18/imran-khan- pakistans-jailed-ex-leader-uses-ai-deepfake-to-address-o...

2024
[11]

A Review on Generative Adversarial Networks: Algorithms, Theory, and Applications,

J. Gui, Z. Sun, Y. Wen, D. Tao, and J. Ye, “A Review on Generative Adversarial Networks: Algorithms, Theory, and Applications,” IEEE Transactions on Knowledge and Data Engineering, vol. 35, no. 4, pp. 3313–3332, Apr. 2023, doi: 10.1109/TKDE.2021.3130191

2023
[12]

Conditional Generative Adversarial Nets,

M. Mirza and S. Osindero, “Conditional Generative Adversarial Nets,” Nov. 06, 2014, arXiv: arXiv:1411.1784. doi: 10.48550/arXiv.1411.1784

2014 doi arXiv
[13]

Generative Adversarial Networks: An Overview,

A. Creswell, T. White, V. Dumoulin, K. Arulkumaran, B. Sengupta, and A. A. Bharath, “Generative Adversarial Networks: An Overview,” IEEE Signal Processing Magazine , vol. 35, no. 1, pp. 53 –65, Jan. 2018, doi: 10.1109/MSP.2017.2765202

2018
[14]

A Style -Based Generator Architecture for Generative Adversarial Networks,

T. Karras, S. Laine, and T. Aila, “A Style -Based Generator Architecture for Generative Adversarial Networks,” Mar. 29, 2019, arXiv: arXiv:1812.04948. doi: 10.48550/arXiv.1812.04948

2019 doi arXiv
[15]

A New Approach to in Ensemble Method for Deepfake Detection,

S. Atas and M. Karakose, “A New Approach to in Ensemble Method for Deepfake Detection,” in 2023 4th International Conference on Data Analytics for Business and Industry (ICDABI) , Oct. 2023, pp. 201–204. doi: 10.1109/ICDABI60145.2023.10629338

2023
[16]

DeepSight: Enhancing Deepfake Image Detection and Classification through Ensemble and Deep Learning Techniques,

T. Manju and S. Kalarani, “DeepSight: Enhancing Deepfake Image Detection and Classification through Ensemble and Deep Learning Techniques,” in 2024 5th International Conference on Image Processing and Capsule Networks (ICIPCN) , Jul. 2024, pp. 28 –35. doi: 10.1109/ICIPCN63822....

2024
[17]

Unmasking Deepfakes - Harnessing the Potential of 2D and 3D Convolutional Neural Network Ensembles,

A. Bakliwal, A. D. Joshi, N. Deo, and S. Sawant, “Unmasking Deepfakes - Harnessing the Potential of 2D and 3D Convolutional Neural Network Ensembles,” in 2024 5th International Conference on Innovative Trends in Information Technology (ICITIIT) , Mar. 2024, pp. 1–6. doi: 10.11...

2024
[18]

EfficientNetB0 Ensemble Model for Unified Deepfakes Detection,

S. A. Minhas, S. Mushtaq, and A. Javed, “EfficientNetB0 Ensemble Model for Unified Deepfakes Detection,” in 2023 17th International Conference on Open Source Systems and Technologies (ICOSST) , Dec. 2023, pp. 1–5. doi: 10.1109/ICOSST60641.2023.10414228

2023
[19]

Hybrid Transformer Network for Deepfake Detection,

S. A. Khan and D. -T. Dang-Nguyen, “Hybrid Transformer Network for Deepfake Detection,” in Proceedings of the 19th International Conference on Content -based Multimedia Indexing , in CBMI ’22. New York, NY, USA: Association for Computing Machinery, Oct. 2022, pp. 8–14. doi: 10...

2022
[20]

Adapter-Based Incremental Learning for Face Forgery Detection,

C. Gao, Q. Xu, P. Qiao, K. Xu, X. Qian, and Y. Dou, “Adapter-Based Incremental Learning for Face Forgery Detection,” in ICASSP 2024 - 2024 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) , Seoul, Korea, Republic of: IEEE, Apr. 2024, pp. 4690–4...

2024
[21]

SIGMA -DF: Single-Side Guided Meta -Learning for Deepfake Detection,

B. Han, J. Li, W. Ren, M. Luo, J. Liu, and X. Cao, “SIGMA -DF: Single-Side Guided Meta -Learning for Deepfake Detection,” in Proceedings of the 2023 ACM International Conference on Multimedia Retrieval , in ICMR ’23. New York, NY, USA: Association for Computing Machinery, Jun....

2023
[22]

Defending Against Deepfakes with Ensemble Adversarial Perturbation,

W. Guan, Z. He, W. Wang, J. Dong, and B. Peng, “Defending Against Deepfakes with Ensemble Adversarial Perturbation,” in 2022 26th International Conference on Pattern Recognition (ICPR), Aug. 2022, pp. 1952–1958. doi: 10.1109/ICPR56361.2022.9956501

2022
[23]

DeepfakeStack: A Deep Ensemble - based Learning Technique for Deepfake Detection,

Md. S. Rana and A. H. Sung, “DeepfakeStack: A Deep Ensemble - based Learning Technique for Deepfake Detection,” in 2020 7th IEEE International Conference on Cyber Security and Cloud Computing (CSCloud)/2020 6th IEEE International Conference on Edge Computing and Scalable Cloud...

2020
[24]

Robust DeepFake Detection Method based on Ensemble of ViT and CNN,

H. Ha, M. Kim, S. Han, and S. Lee, “Robust DeepFake Detection Method based on Ensemble of ViT and CNN,” in Proceedings of the 38th ACM/SIGAPP Symposium on Applied Computing , in SAC ’23. New York, NY, USA: Association for Computing Machinery, Jun. 2023, pp. 1092–1095. doi: 10....

2023
[25]

ID- Reveal: Identity -aware DeepFake Video Detection,

D. Cozzolino, A. Rössler, J. Thies, M. Nießner, and L. Verdoliva, “ID- Reveal: Identity -aware DeepFake Video Detection,” in 2021 IEEE/CVF International Conference on Computer Vision (ICCV) , Oct. 2021, pp. 15088–15097. doi: 10.1109/ICCV48922.2021.01483

2021
[26]

Xception: Deep Learning with Depthwise Separable Convolutions,

F. Chollet, “Xception: Deep Learning with Depthwise Separable Convolutions,” Apr. 04, 2017, arXiv: arXiv:1610.02357. doi: 10.48550/arXiv.1610.02357

2017 doi arXiv
[27]

Res2Net: A New Multi -scale Backbone Architecture,

S.-H. Gao, M.-M. Cheng, K. Zhao, X.-Y. Zhang, M.-H. Yang, and P. Torr, “Res2Net: A New Multi -scale Backbone Architecture,” IEEE Trans. Pattern Anal. Mach. Intell. , vol. 43, no. 2, pp. 652 –662, Feb. 2021, doi: 10.1109/TPAMI.2019.2938758

2021
[28]

EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks,

M. Tan and Q. V. Le, “EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks,” Sep. 11, 2020, arXiv: arXiv:1905.11946. doi: 10.48550/arXiv.1905.11946

2020 doi arXiv
[29]

DeepFake Detection for Human Face Images and Videos: A Survey | IEEE Journals & Magazine | IEEE Xplore

“DeepFake Detection for Human Face Images and Videos: A Survey | IEEE Journals & Magazine | IEEE Xplore.” Accessed: Nov. 30, 2024. [Online]. Available: https://ieeexplore.ieee.org/abstract/document/9712265

2024
[30]

DeepFake Detection for Human Face Images and Videos: A Survey,

A. Malik, M. Kuribayashi, S. M. Abdullahi, and A. N. Khan, “DeepFake Detection for Human Face Images and Videos: A Survey,” IEEE Access , vol. 10, pp. 18757 –18775, 2022, doi: 10.1109/ACCESS.2022.3151186

2022
[31]

Celeb-DF: A Large-scale Challenging Dataset for DeepFake Forensics,

Y. Li, X. Yang, P. Sun, H. Qi, and S. Lyu, “Celeb-DF: A Large-scale Challenging Dataset for DeepFake Forensics,” Mar. 16, 2020, arXiv: arXiv:1909.12962. doi: 10.48550/arXiv.1909.12962

2020 doi arXiv
[32]

FaceForensics++: Learning to Detect Manipulated Facial Images,

A. Rössler, D. Cozzolino, L. Verdoliva, C. Riess, J. Thies, and M. Nießner, “FaceForensics++: Learning to Detect Manipulated Facial Images,” Aug. 26, 2019, arXiv: arXiv:1901.08971. doi: 10.48550/arXiv.1901.08971

2019 doi arXiv
[33]

Deepfake Detection: Analyzing Model Generalization Across Architectures, Datasets, and Pre - Training Paradigms,

S. A. Khan and D.-T. Dang-Nguyen, “Deepfake Detection: Analyzing Model Generalization Across Architectures, Datasets, and Pre - Training Paradigms,” IEEE Access, vol. 12, pp. 1880–1908, 2024, doi: 10.1109/ACCESS.2023.3348450

1908
[34]

An Analysis of Recent Advances in Deepfake Image Detection in an Evolving Threat Landscape,

S. M. Abdullah et al., “An Analysis of Recent Advances in Deepfake Image Detection in an Evolving Threat Landscape,” in 2024 IEEE Symposium on Security and Privacy (SP), May 2024, pp. 91–109. doi: 10.1109/SP54263.2024.00194

2024
[35]

C++, Python

dlib: A toolkit for making real world machine learning and data analysis applications . C++, Python. Accessed: Dec. 13, 2024. [MacOS :: MacOS X, Microsoft, Microsoft :: Windows, POSIX, POSIX :: Linux]. Available: https://github.com/davisking/dlib

2024
[36]

M. D. Bloice, mdbloice/Augmentor. (Dec. 03, 2024). Python. Accessed: Dec. 04, 2024. [Online]. Available: https://github.com/mdbloice/Augmentor

2024
[37]

Wightman, rwightman/timm

R. Wightman, rwightman/timm. (Nov. 26, 2024). Python. Accessed: Nov. 30, 2024. [Online]. Available: https://github.com/rwightman/timm

2024
[38]

How to maximize CPU <==> GPU memory transfer speeds?,

“How to maximize CPU <==> GPU memory transfer speeds?,” PyTorch Forums. Accessed: Dec. 14, 2024. [Online]. Available: https://discuss.pytorch.org/t/how-to-maximize-cpu-gpu-memory- transfer-speeds/173855

2024
[2024]

Available: https://blog.whatsapp.com/talk-to-meta-ai- on-whatsapp

[Online]. Available: https://blog.whatsapp.com/talk-to-meta-ai- on-whatsapp

Pith tools

Reviewed August 16, 2026 · model on record in the stance chip above.

[1] [1]

Talk to Meta AI on WhatsApp,

“Talk to Meta AI on WhatsApp,” WhatsApp.com. Accessed: Dec. 04,

work page

[2] [2]

Say Hi to My AI

“Say Hi to My AI.” Accessed: Dec. 04, 2024. [Online]. Available: https://newsroom.snap.com/say-hi-to-my-ai

work page 2024

[3] [3]

The Creation and Detection of Deepfakes: A Survey,

Y. Mirsky and W. Lee, “The Creation and Detection of Deepfakes: A Survey,” ACM Comput. Surv., vol. 54, no. 1, p. 7:1 -7:41, Jan. 2021, doi: 10.1145/3425780

doi:10.1145/3425780 2021

[4] [4]

NSA, U.S. Federal Agencies Advise on Deepfake Threats,

“NSA, U.S. Federal Agencies Advise on Deepfake Threats,” National Security Agency/Central Security Service. Accessed: Nov. 30, 2024. [Online]. Available: https://www.nsa.gov/Press -Room/Press- Releases-Statements/Press-Release-View/Article/3523329/nsa-us- federal-agencies-advise-on-deepfake- threats/http%3A%2F%2Fwww.nsa.gov%2FPress-Room%2FPress- Releases-...

arXiv 2024

[5] [5]

Battle of Deep Fakes: Artificial Intelligence Set to Become a Major Threat to the Individual and National Security,

A. Ali, K. F. Khan Ghouri, H. Naseem, T. R. Soomro, W. Mansoor, and A. M. Momani, “Battle of Deep Fakes: Artificial Intelligence Set to Become a Major Threat to the Individual and National Security,” in 2022 International Conference on Cyber Resilience (ICCR) , Oct. 2022, pp. 1–5. doi: 10.1109/ICCR56254.2022.9995821

arXiv 2022

[6] [6]

Deepfakes and the Dangers to National Security and Defence - RSIS

“Deepfakes and the Dangers to National Security and Defence - RSIS.” Accessed: Dec. 04, 2024. [Online]. Available: https://www.rsis.edu.sg/rsis-publication/rsis/deepfakes-and-the- Architectures FF++ CelebDF-v2 AUC Accuracy Precision Recall F1 AUC Accuracy Precision Recall F1 Xception 99.25% 97.04% 97.74% 98.60% 98.17% 100.00% 99.90% 99.91% 99.94% 99.92% R...

work page arXiv 2024

[7] [7]

Protecting World Leaders Against Deep Fakes,

S. Agarwal, H. Farid, Y. Gu, M. He, K. Nagano, and H. Li, “Protecting World Leaders Against Deep Fakes,” in IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Workshops, IEEE, 2019, pp. 38 –45. Accessed: Nov. 30, 2024. [Online]. Available: https://openaccess.thecvf.com/content_CVPRW_2019/html/Media_F orensics/Agarwal_Protecting_World_Le...

work page 2019

[8] [8]

Deepfake in Movies | Facts, Fiction & Future of Film Industry,

teresa_myers, “Deepfake in Movies | Facts, Fiction & Future of Film Industry,” Facia.ai. Accessed: Dec. 04, 2024. [Online]. Available: https://facia.ai/blog/deepfake-in-movies-facts-fiction-future-of-film- industry/

work page 2024