Pith. sign in

REVIEW 4 major objections 5 minor 46 references

A multimodal fake news detector that calibrates text–image coherence, normalizes domain statistics, and distills teacher knowledge into domain-expert students achieves the highest reported accuracy while cutting cross-domain error disparity

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

EGMD reduces cross-domain bias in multimodal fake-news detection by calibrating input coherence, aligning domain statistics with a normalized expert teacher, and distilling into prototype-anchored student branches.

T0 review reviewed 2026-08-01 challenge →

load-bearing objection The architecture results hold up on the public benchmarks, but the Weibo_Balanced control doesn't isolate imbalance, so the headline 57.3% claim is the weakest part of the paper. the 4 major comments →

arxiv 2607.26555 v1 pith:4YC6WVDI submitted 2026-07-29 cs.CL

Where Detectors Fail: Closing the Tail-Domain Gap with Expert-Guided Mutual Distillation

Xuan Feng , Guihong Liu , Tianlong Gu , Shuai Zhao , Xuemin Wang , Chenzhong Bin , Yang Liu , Bo An This is my paper
classification cs.CL
keywords multimodal fake news detectiondomain biasdomain generalizationknowledge distillationmixture of expertssemantic misalignmentimbalanced datafairness metrics
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

The reading

Multimodal fake news detectors often look accurate on average while failing badly on underrepresented domains, because they learn domain-specific shortcuts and trust text–image pairs that do not actually agree. The paper argues that this failure can be addressed by deciding what evidence to trust at three points in the pipeline: before fusion, in the representation space, and at the decision boundary. It proposes EGMD, a teacher–student framework whose input-level calibration gates fusion on pair coherence, whose expert-guided teacher aligns domain statistics and routes residual patterns through specialized experts, and whose prototype-anchored students reach consensus through mutual distillation. Across four datasets in Chinese and English, EGMD reports the highest overall accuracy while reducing cross-domain error disparity by up to 57.3% and raising worst-domain accuracy. The paper also constructs Weibo_Balanced to show that balancing domain volumes further cuts EGMD's disparity by 37.7%, supporting domain imbalance as a major driver of the bias.

Core claim

On its own terms, the paper's central claim is that cross-domain failure in multimodal fake news detection follows a single error path—unreliable cross-modal evidence contaminates representations, entangles with domain statistics, and biases decisions—and that intervening at each stage removes the bias. EGMD implements this as three dependent innovations: input-level calibration encodes pair-level coherence as a shared multiplicative gain applied before fusion; representation-level debiasing standardizes domain-conditioned statistics (DDN) and uses a domain-conditioned mixture-of-experts to concentrate domain-specific patterns; decision-level generalization anchors domain-specific student br

What carries the argument

The carrying mechanism is a three-stage teacher–student pipeline. Stage one, input-level calibration, computes a CLIP-based coherence scalar between text and image and applies it as a shared gain before fusion, so incompatible pairs are never fused blindly. Stage two, the expert-guided teacher, uses Dynamic Domain-specific Normalization (DDN)—batch standardization with domain-conditioned affine parameters—followed by a domain-conditioned Mixture-of-Experts that routes residual domain patterns into specialized experts. Stage three, the mutual distillation student, anchors each domain branch to a moving prototype centroid, enforces pairwise KL consensus between branches, and distills the teach

Load-bearing premise

The load-bearing premise is that Weibo_Balanced isolates the effect of imbalance: the newly crawled, expert-annotated instances added to underrepresented domains are assumed to differ only in volume, not in topic, recency, or annotation style; if that fails, the 37.7% reduction is a distribution-shift result rather than evidence about imbalance.

What would settle it

Construct a balanced version of Weibo21 by random subsampling alone (no new data collection) and retrain EGMD under the same protocol; if Total no longer drops by roughly 37.7% relative to the imbalanced benchmark, the claim that domain imbalance itself drives the disparity is falsified.

Watch this falsifier. Get emailed when new claim-graph text bears on it.

If this is right

  • Mean accuracy should not be read alone; reporting worst-domain accuracy and cross-domain error disparity alongside it exposes deployment risk that averages hide.
  • Imbalanced domain volume is a major but not sufficient cause of domain bias: balancing data helps only if the model is built to resist domain-specific shortcuts.
  • Detectors can be made more uniform across domains without sacrificing aggregate accuracy; EGMD's worst-domain accuracy improves on all three public benchmarks.
  • A single lightweight student branch suffices at inference, with learned prototypes standing in for domain labels when they are unavailable.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • A strict random-subsample version of Weibo_Balanced (no new crawl) would tell whether the 37.7% drop is pure volume balancing or partly content and topic shift.
  • The same three-stage recipe—coherence gating, domain-normalized experts, and consensus distillation—should transfer to other multi-domain classification tasks with tail groups, such as hate-speech or medical-triage detection.
  • Because the input-level gain is strictly amplificatory, the method is designed to tolerate benign stock imagery; an attenuating variant is a natural ablation to test whether suppression ever helps.
  • Prototype-based routing is a closed-set rule; extending EGMD to genuinely novel domains would require an open-set routing mechanism.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 5 minor

Summary. The paper proposes Expert-Guided Mutual Distillation (EGMD), a three-stage teacher–student framework for multimodal fake news detection that targets domain bias and semantic misalignment. Input-level calibration encodes CLIP-based text–image coherence as a shared gain before fusion; representation-level debiasing uses Dynamic Domain-specific Normalization and a domain-conditioned MoE; decision-level generalization uses prototype-anchored mutual distillation and dual-channel distillation. The paper evaluates on Weibo, Weibo21, FineFake, and a newly constructed Weibo_Balanced benchmark, reporting aggregate accuracy, worst-domain accuracy, and a cross-domain disparity score Total. The headline claims are state-of-the-art accuracy with the lowest disparity, and that reducing domain imbalance from 11.08× to 1.27× lowers EGMD's Total by 37.7%, supporting imbalance as a major driver of domain bias.

Significance. The empirical architecture is clearly specified and the per-domain tables are the strongest part of the paper. On Weibo, Weibo21, and FineFake, EGMD reports the highest worst-domain accuracy and the lowest Total disparity, and the ablation study, progressive component analysis, and missing-domain-label robustness experiments support the internal consistency of the design. The use of external public benchmarks and rerunning baselines under identical splits are also strengths. If the causal claim about domain imbalance were properly supported, the contribution would be valuable for the fake-news-detection community. However, the causal claim and the 'up to 57.3%' headline rest on Weibo_Balanced, whose construction is confounded, and the absence of variance reporting makes the significance claims unverifiable. These issues are fixable but require substantive revision.

major comments (4)
  1. [§3.2, Appendix G.2, §5.4] The claim that Weibo_Balanced 'isolates the effect of imbalance' is not supported. The benchmark is built by targeted keyword-based crawling of underrepresented domains and downsampling Society, so it changes content selection, topic mix, annotation time, and label distribution, not merely the domain counts n_d. Consequently, the 37.7% Total reduction (0.5415→0.3376) reported in §5.4 cannot be attributed to reduced imbalance; it may reflect distribution shift of the new instances. The 'up to 57.3%' abstract claim is also computed against this benchmark. A proper control would re-sample the original Weibo21 instances to balance n_d (or use stratified reweighting) while keeping the same instance population, or the claims must be reframed as 'on a separately collected balanced corpus' rather than 'isolating' imbalance.
  2. [§5.1, Tables 2–3] The paper states that results are averaged over five independent runs and that a t-test against the runner-up determines significance, but no standard deviations, confidence intervals, or per-run values are reported anywhere. This is especially problematic for claims that are close, e.g., Weibo21 overall accuracy 91.63% vs. 91.58% for DAMMFND; without variance, the † markers cannot be evaluated. Please report std/CI for all main tables, or state clearly why significance testing is possible without them.
  3. [§5.6, Figure 6] The hyperparameters α, β, τ_c, and λ are selected from Figure 6, but the text does not state which split or dataset is used for this selection, nor whether the reported test numbers are the same runs used for selection. If the same benchmark is used for both tuning and final reporting, the results are susceptible to selection bias. Also, τ_f and T_d are said to be fixed but their values are not reported in the main text or Appendix A; please specify them.
  4. [§4.1, Eq. (2)–(3)] The input-level calibration gate is strictly amplificatory: 1+s ∈ (1,2), so low-coherence pairs are never attenuated. The authors acknowledge this in the text, and it is a defensible conservative choice, but it weakens the abstract's framing that the model 'learns what evidence to trust' at the input level. This is not a fatal flaw, but the contribution phrasing should be aligned with what the component actually does.
minor comments (5)
  1. [§5.1 / Tables 2–3] The domain header 'Int./Dis.' in Table 2 is ambiguous; define whether this column aggregates the International and Disaster domains and why.
  2. [Appendix G] No availability statement or URL is given for Weibo_Balanced, nor for the code. Releasing the constructed benchmark and code would materially help reproducibility.
  3. [Appendix D] PFS is explicitly preference-dependent and auxiliary; this is transparently stated. However, since θ=0.8 is arbitrary, I suggest moving PFS to a supplementary role in the main text and relying on Accuracy, Worst, and Total, as the paper already does.
  4. [Appendix C.3] The exclusion of ADOSE and MemiMoE-FND is reasonable due to protocol differences, but the sentence 'the required supervision or reported outputs are not directly comparable' should also note that ADOSE uses 10% target labels, because this explains the exclusion in one line.
  5. [References] There are self-citations to DTDBD (Li et al. 2024) and C2PO (Feng et al. 2026); these are appropriate as related work and do not appear to influence the benchmark results.

Circularity Check

0 steps flagged

No significant circularity: EGMD is an empirical architecture and its headline results are comparisons on public benchmarks, not consequences of its own definitions or self-citations.

full rationale

EGMD is presented as an architecture, not as a theorem derivation. The central reported numbers—SOTA overall/worst-domain accuracy and reduced Total on Weibo, Weibo21, and FineFake, plus the Weibo→Weibo21 transfer result—are measured against public benchmarks with baselines rerun under identical splits, so they do not reduce to the paper's definitions. The training objective L_total includes mutual-distillation and logit/feature-distillation terms, but the headline metric Total=FNED+FPED is computed from per-domain error rates rather than from the KL/InfoNCE terms being optimized; the low Total is therefore an empirical outcome, not an identity. The constructed Weibo_Balanced benchmark and the PFS scalar are accompanied by explicit caveats (Appendix D: PFS is 'only a supplementary summary rather than as standalone evidence of superiority'; Appendix J: t-SNE is 'treated as qualitative supporting evidence'; §1: the semantic-mismatch figure 'does not establish that mismatch alone determines veracity' and main claims remain 'grounded in the naturally labeled benchmark evaluations'). These caveats reduce any risk of a self-definitional equivalence. Self-citations to DTDBD (Li et al. 2024) and C2PO (Feng et al. 2026) appear only for background and metric definition; no load-bearing argument, uniqueness theorem, or adopted ansatz depends on them. The Weibo_Balanced causal claim has a real methodological confound (§3.2/Appendix G.2)—targeted crawling changes content, topic mix, and annotation in addition to n_d—but that is a validity/correctness concern, not circularity: the 37.7% figure is an empirical comparison across datasets, not an equation that equals its own input. No circular step can be exhibited with the required quote-and-reduction specificity.

Axiom & Free-Parameter Ledger

6 free parameters · 5 axioms · 2 invented entities

The method rests on standard supervised-learning assumptions plus several domain-specific premises. The most fragile premise is that Weibo_Balanced isolates imbalance: because the balanced set adds newly crawled instances rather than re-sampling the original population, the causal interpretation of the 37.7% Total reduction is insecure. Hyperparameters α, β, τ_c, λ were tuned via sensitivity analysis and then reused across datasets.

free parameters (6)
  • Mutual-distillation weight α = 0.3
    Chosen from sensitivity analysis in §5.6; affects accuracy/disparity trade-off, with α=0.5 increasing Total without improving accuracy.
  • Dual-distillation weight β = 0.3
    Chosen from §5.6; method is comparatively robust to β (Total varies 0.35–0.38).
  • Coherence temperature τ_c = 0.1
    Selected in §5.6 as the optimum for Accuracy, PFS, and Total; larger temperatures weaken the balance.
  • Prototype momentum λ = 0.9
    Selected in §5.6; λ=0.9 maximizes Accuracy/PFS and minimizes disparity.
  • PFS preference weight θ = 0.8
    Hand-chosen scalar in Eq. (12) to prioritize accuracy; the paper explicitly calls it preference-dependent and auxiliary.
  • Distillation temperatures τ_f and T_d = not stated
    Described as fixed in §5.6 but values are not reported; a reproducibility gap.
axioms (5)
  • domain assumption Ground-truth domain labels are available at training time and the inference taxonomy is closed to the K training domains.
    Invoked in §4.4 and Appendix H.2; the teacher's DDN, MoE routing, and student branches all rely on domain identity, with prototype-based routing as only a fallback.
  • domain assumption CLIP cosine similarity between text and image embeddings is a valid scalar proxy for pair-level semantic coherence.
    Used in Eq. (2) to define the shared gain s; if CLIP similarity is unreliable for these news images, the input-level calibration injects noise.
  • domain assumption Batch-statistic normalization plus domain-conditioned affine parameters (DDN) aligns domain distributions without destroying veracity information.
    Eq. (4) assumes the residual after normalization is safe to route through experts; the paper itself notes DDN may attenuate magnitude information.
  • domain assumption Forcing all domain-specific student branches to agree via pairwise KL divergence improves generalization rather than propagating a biased consensus.
    Eq. (8) enforces consensus on P(Y); if the shared teacher is biased in some domains, mutual distillation can distribute that bias.
  • domain assumption Newly collected, expert-annotated instances in Weibo_Balanced are comparable to original Weibo21 instances except for domain volume.
    Appendix G.2 describes targeted keyword-based collection; if the added instances come from different events or time periods, balancing is confounded with content shift.
invented entities (2)
  • Domain-specific prototype memories p_k no independent evidence
    purpose: Moving centroids that geometrically anchor each student branch's feature distribution (Eq. 7).
    No standalone falsifiable prediction; their effect is only observable through downstream accuracy and disparity numbers.
  • Expert-Guided Teacher with Dynamic Domain-specific Normalization no independent evidence
    purpose: Training-only teacher module that aligns domain statistics and routes residual patterns through domain-conditioned experts.
    Architectural construct with no external measurable signature beyond the benchmark evaluations.

reviewed 2026-08-01 · how reviews work

0 comments
Cite this review

Pith. "Pith review of Where Detectors Fail: Closing the Tail-Domain Gap with Expert-Guided Mutual Distillation." pith.science (2026). https://pith.science/paper/4YC6WVDI

@misc{pith2026260726555,
  author       = {Pith},
  title        = {Pith review of: Where Detectors Fail: Closing the Tail-Domain Gap with Expert-Guided Mutual Distillation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/4YC6WVDI}},
  note         = {Machine review of arXiv:2607.26555}
}
Share X Bluesky LinkedIn Reddit HN
read the original abstract

Multimodal fake news detectors often generalize poorly across domains because they learn to trust unreliable evidence: domain-specific shortcuts amplified by imbalanced data and semantically inconsistent text-image pairs that make cross-modal evidence unreliable. We propose Expert-Guided Mutual Distillation (EGMD), which learns what evidence to trust across the prediction pipeline. At the input level, input-level calibration encodes pair-level coherence as a shared gain before fusion. At the representation level, an expert-guided teacher aligns domain statistics and encourages domain-specific patterns to concentrate in specialized experts. At the decision level, prototype-anchored domain-specific students use mutual learning and dual-channel distillation to inherit the teacher's feature geometry and calibrated predictions while discouraging local domain priors. We further construct Weibo_Balanced, a domain-balanced benchmark that isolates the effect of imbalance on generalization. Across four datasets in two languages, EGMD achieves state-of-the-art accuracy while reducing domain bias by up to 57.3%.

Figures

Figures reproduced from arXiv: 2607.26555 by Bo An, Chenzhong Bin, Guihong Liu, Shuai Zhao, Tianlong Gu, Xuan Feng, Xuemin Wang, Yang Liu.

Figure 1
Figure 1. Figure 1: Illustration of Semantic Misalignment. Moti￾vated by recurring prediction errors observed in our error analysis, we construct a clear example by pairing the same Health text with (i) its matched image and (ii) an unrelated Politics image. Bars report P(Fake), illustrating how model confidence changes under pronounced cross-modal conflict. design, while the main performance claims remain grounded in the nat… view at source ↗
Figure 2
Figure 2. Figure 2: The EGMD framework comprises three stages: Input-Level Calibration (§4.1), Representation-Level Debiasing via Dynamic Domain-specific Normalization and Mixture-of-Experts (§4.2), and Decision-Level Generalization via a prototype￾anchored Mutual Distillation Student (§4.3). Teacher-student transfer uses feature- and logit-level distillation losses (Lfeat, Llogit). Step 1: Aligning Domain Distributions via D… view at source ↗
Figure 3
Figure 3. Figure 3: Per-domain accuracy ranges on Weibo21 and Fine [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figure 5
Figure 5. Figure 5: Ablation study on Weibo and Weibo21. DDN: Dy [PITH_FULL_IMAGE:figures/full_fig_p007_5.png] view at source ↗
Figure 4
Figure 4. Figure 4: Overall Accuracy and PFS on (a) cross-dataset [PITH_FULL_IMAGE:figures/full_fig_p007_4.png] view at source ↗
Figure 7
Figure 7. Figure 7: Performance comparison between head and tail [PITH_FULL_IMAGE:figures/full_fig_p012_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: t-SNE visualization of feature distributions on [PITH_FULL_IMAGE:figures/full_fig_p014_8.png] view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

46 extracted references · 2 linked inside Pith

  1. [1]

    Cai, Y.; Xi, M.; Shang, Y.; and Yin, J. 2023. Exploring High-Correlation Source Domain Information for Multi-Source Domain Adaptation in Semantic Segmentation. In Proceedings of the 31st ACM International Conference on Multimedia, 2148--2158

    2023

  2. [2]

    Chen, Y.; Shi, W.; Li, M.; Cui, Y.; Li, J.; Zhang, R.; Chen, H.; Guo, H.; Di, S.; Liu, Z.; Zhu, J.; and Xu, J. 2026. Active Multi-Source Domain Adaptation for Multimodal Fake News Detection. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 40, 20363--20371

    2026

  3. [3]

    Cui, X.; Lu, W.; Tong, Y.; Li, Y.; and Zhao, Z. 2025 a . Diffusion-Based Multi-Modal Synergy Interest Network for Click-Through Rate Prediction. In Proceedings of the 48th International ACM SIGIR Conference on Research and Development in Information Retrieval, 581--591

    2025

  4. [4]

    Cui, X.; Lu, W.; Tong, Y.; Li, Y.; and Zhao, Z. 2025 b . Multi-Modal Multi-Behavior Sequential Recommendation with Conditional Diffusion-Based Feature Denoising. In Proceedings of the 48th International ACM SIGIR Conference on Research and Development in Information Retrieval, 1593--1602

    2025

  5. [5]

    Devlin, J.; Chang, M.-W.; Lee, K.; and Toutanova, K. 2019. BERT : Pre-Training of Deep Bidirectional Transformers for Language Understanding. In Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, 4171--4186

    2019

  6. [6]

    Dixon, L.; Li, J.; Sorensen, J.; Thain, N.; and Vasserman, L. 2018. Measuring and Mitigating Unintended Bias in Text Classification. In Proceedings of the 2018 AAAI/ACM Conference on AI, Ethics, and Society, 67--73

    2018

  7. [7]

    S.; and Sun, L

    Dou, Y.; Shu, K.; Xia, C.; Yu, P. S.; and Sun, L. 2021. User Preference-Aware Fake News Detection. In Proceedings of the 44th International ACM SIGIR Conference on Research and Development in Information Retrieval, 2051--2055

    2021

  8. [8]

    Feng, X.; An, B.; Gu, T.; Chang, L.; Hao, F.; Yu, P.; and Zhao, S. 2026. C2PO : Diagnosing and Disentangling Bias Shortcuts in LLMs . In Findings of the Association for Computational Linguistics: ACL 2026, 24496--24515

    2026

  9. [9]

    Guo, H.; Ma, Z.; Zeng, Z.; Luo, M.; Zeng, W.; Tang, J.; and Zhao, X. 2025. Each Fake News Is Fake in Its Own Way: An Attribution Multi-Granularity Benchmark for Multimodal Fake News Detection. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 39, 228--236

    2025

  10. [10]

    He, K.; Chen, X.; Xie, S.; Li, Y.; Doll \'a r, P.; and Girshick, R. 2022. Masked Autoencoders Are Scalable Vision Learners. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 16000--16009

    2022

  11. [11]

    Lao, A.; Zhang, Q.; Shi, C.; Cao, L.; Yi, K.; Hu, L.; and Miao, D. 2024. Frequency Spectrum Is More Effective for Multimodal Representation and Fusion: A Multimodal Spectrum Rumor Detector. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 38, 18426--18434

    2024

  12. [12]

    Li, J.; Feng, X.; Gu, T.; and Chang, L. 2024. Dual-Teacher De-biasing Distillation Framework for Multi-Domain Fake News Detection. In IEEE 40th International Conference on Data Engineering, 3627--3639

    2024

  13. [13]

    Li, Y.; Dong, J.; Dong, Z.; Yang, C.; An, Z.; and Xu, Y. 2025 a . SRKD : Towards Efficient 3D Point Cloud Segmentation via Structure- and Relation-Aware Knowledge Distillation. arXiv preprint arXiv:2506.17290

    Pith/arXiv arXiv 2025

  14. [14]

    Li, Y.; Lin, X.; Zhang, K.; Yang, C.; Guo, Z.; Gou, J.; and Li, Y. 2025 b . FedKD-Hybrid : Federated Hybrid Knowledge Distillation for Lithography Hotspot Detection. arXiv preprint arXiv:2501.04066

    Pith/arXiv arXiv 2025

  15. [15]

    Li, Y.; Long, Q.; Zhou, Y.; Zhang, R.; Ning, Z.; Zhu, Z.; Zhou, Y.; Wang, X.; and Xiao, M. 2025 c . ComAE : Comprehensive Attribute Exploration for Zero-Shot Hashing. In Proceedings of the 2025 International Conference on Multimedia Retrieval, 733--742

    2025

  16. [16]

    Li, Y.; Yang, C.; Zeng, H.; Dong, Z.; An, Z.; Xu, Y.; Tian, Y.; and Wu, H. 2025 d . Frequency-Aligned Knowledge Distillation for Lightweight Spatiotemporal Forecasting. In Proceedings of the IEEE/CVF International Conference on Computer Vision, 7262--7272

    2025

  17. [17]

    Liu, H.; Jin, W.; Karimi, H.; Liu, Z.; and Tang, J. 2021. The Authors Matter: Understanding and Mitigating Implicit Bias in Deep Text Classification. In Findings of the Association for Computational Linguistics: ACL-IJCNLP 2021, 74--85

    2021

  18. [18]

    Liu, H.; Wang, W.; Li, H.; and Li, H. 2024. TELLER : A Trustworthy Framework for Explainable, Generalizable and Controllable Fake News Detection. In Findings of the Association for Computational Linguistics: ACL 2024, 15556--15583

    2024

  19. [19]

    Liu, Y.; Liu, Y.; Li, Z.; Yao, R.; Zhang, Y.; and Wang, D. 2025. Modality Interactive Mixture-of-Experts for Fake News Detection. In Proceedings of the ACM on Web Conference, 5139--5150

    2025

  20. [20]

    Lu, W.; Tong, Y.; and Ye, Z. 2025. DAMMFND : Domain-Aware Multimodal Multi-View Fake News Detection. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 39, 559--567

    2025

  21. [21]

    Ma, Z.; Luo, M.; Guo, H.; Zeng, Z.; Hao, Y.; and Zhao, X. 2024. Event-RADAR : Event-Driven Multi-View Learning for Multimodal Fake News Detection. In Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics, 5809--5821

    2024

  22. [22]

    Meng, Y.; Wang, H.; Zhao, J.; Sun, Y.; and Shao, M. 2026. Dynamic Hierarchical Memory Improved Mixture-of-Experts for Multimodal Fake News Detection. Information Processing & Management, 63(2): 104479

    2026

  23. [23]

    V.; and Liu, H

    Mosallanezhad, A.; Karami, M.; Shu, K.; Mancenido, M. V.; and Liu, H. 2022. Domain adaptive fake news detection via reinforcement learning. In Proceedings of the ACM Web Conference, 3632--3640

    2022

  24. [24]

    Nan, Q.; Cao, J.; Zhu, Y.; Wang, Y.; and Li, J. 2021. MDFEND : Multi-Domain Fake News Detection. In Proceedings of the 30th ACM International Conference on Information & Knowledge Management, 3343--3347

    2021

  25. [25]

    Peng, L.; Jian, S.; Li, D.; and Shen, S. 2023. MRML : Multimodal Rumor Detection by Deep Metric Learning. In IEEE International Conference on Acoustics, Speech and Signal Processing, 1--5

    2023

  26. [26]

    Qi, P.; Cao, J.; Li, X.; Liu, H.; Sheng, Q.; Mi, X.; He, Q.; Lv, Y.; Guo, C.; and Yu, Y. 2021. Improving Fake News Detection by Using an Entity-Enhanced Framework to Fuse Diverse Multimodal Clues. In Proceedings of the 29th ACM international conference on multimedia, 1212--1220

    2021

  27. [27]

    Qiao, J.; Li, X.; Gao, C.; Wu, L.; Feng, J.; and Wang, Z. 2025. Improving Multimodal Fake News Detection by Leveraging Cross-Modal Content Correlation. Information Processing & Management, 62(5): 104120

    2025

  28. [28]

    W.; Hallacy, C.; Ramesh, A.; Goh, G.; Agarwal, S.; Sastry, G.; Askell, A.; Mishkin, P.; Clark, J.; et al

    Radford, A.; Kim, J. W.; Hallacy, C.; Ramesh, A.; Goh, G.; Agarwal, S.; Sastry, G.; Askell, A.; Mishkin, P.; Clark, J.; et al. 2021. Learning Transferable Visual Models from Natural Language Supervision. In International Conference on Machine Learning, 8748--8763. PmLR

    2021

  29. [29]

    Shen, L.; Long, Y.; Cai, X.; Razzak, I.; Chen, G.; Liu, K.; and Jameel, S. 2025. GAMED : Knowledge-Adaptive Multi-Experts Decoupling for Multimodal Fake News Detection. In Proceedings of the Eighteenth ACM International Conference on Web Search and Data Mining, 586--595

    2025

  30. [30]

    Sun, J.; Chen, C.; Hou, C.; Wu, Y.; and Yuan, X. 2025. Multimodal Taylor Series Network for Misinformation Detection. In Proceedings of the ACM on Web Conference, 2540--2548

    2025

  31. [31]

    Tong, Y.; Lu, W.; Zhao, Z.; Lai, S.; and Shi, T. 2024. MMDFND : Multi-Modal Multi-Domain Fake News Detection. In Proceedings of the 32nd ACM International Conference on Multimedia, 1178--1186

    2024

  32. [32]

    van der Maaten, L.; and Hinton, G. 2008. Visualizing Data using t-SNE . Journal of Machine Learning Research, 9(86): 2579--2605

    2008

  33. [33]

    Wang, L.; Zhang, C.; Xu, H.; Xu, Y.; Xu, X.; and Wang, S. 2023. Cross-Modal Contrastive Learning for Multimodal Fake News Detection. In Proceedings of the 31st ACM International Conference on Multimedia, 5696--5704

    2023

  34. [34]

    Wang, Q.; Yan, K.; and Ding, S. 2024. Bilateral Adaptive Cross-Modal Fusion Prompt Learning for CLIP . In Proceedings of the 32nd ACM International Conference on Multimedia, 9001--9009

    2024

  35. [35]

    Wang, Y.; Ma, F.; Jin, Z.; Yuan, Y.; Xun, G.; Jha, K.; Su, L.; and Gao, J. 2018. EANN : Event adversarial neural networks for multi-modal fake news detection. In Proceedings of the 24th Acm SIGKDD International Conference on Knowledge Discovery & Data Mining, 849--857

    2018

  36. [36]

    Yang, R.; Gao, W.; Ma, J.; Lin, H.; and Wang, B. 2024. Reinforcement Tuning for Detecting Stances and Debunking Rumors Jointly with Large Language Models. In Findings of the Association for Computational Linguistics: ACL 2024, 13423--13439

    2024

  37. [37]

    Ying, Q.; Hu, X.; Zhou, Y.; Qian, Z.; Zeng, D.; and Ge, S. 2023. Bootstrapping Multi-View Representations for Fake News Detection. In Proceedings of the AAAI conference on Artificial Intelligence, volume 37, 5384--5392

    2023

  38. [38]

    Zeng, Z.; Luo, M.; Kong, X.; Liu, H.; Guo, H.; Yang, H.; Ma, Z.; and Zhao, X. 2024. Mitigating World Biases: A Multimodal Multi-View Debiasing Framework for Fake News Video Detection. In Proceedings of the 32nd ACM international conference on multimedia, 6492--6500

    2024

  39. [39]

    S.; and Li, C

    Zhang, L.; Zhang, X.; Zhou, Z.; Zhang, X.; Yu, P. S.; and Li, C. 2025. Knowledge-Aware Multimodal Pre-Training for Fake News Detection. Information Fusion, 114: 102715

    2025

  40. [40]

    A.; Pan, F.; and Wen, J

    Zhao, S.; Jia, M.; Tuan, L. A.; Pan, F.; and Wen, J. 2024. Universal Vulnerabilities in Large Language Models: Backdoor Attacks for In-context Learning. In Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing, 11507--11522

    2024

  41. [41]

    T.; Jia, Y.; Jia, M.; Yichao, F.; and Tuan, L

    Zhao, S.; Wu, X.; Nguyen, C.-D. T.; Jia, Y.; Jia, M.; Yichao, F.; and Tuan, L. A. 2025. Unlearning Backdoor Attacks for LLMs with Weak-to-Strong Knowledge Distillation. In Findings of the Association for Computational Linguistics: ACL 2025, 4937--4952

    2025

  42. [42]

    Zheng, J.; Zhang, X.; Guo, S.; Wang, Q.; Zang, W.; and Zhang, Y. 2022. MFAN : Multi-Modal Feature-Enhanced Attention Networks for Rumor Detection. In Proceedings of the 31st International Joint Conference on Artificial Intelligence, 2413--2419

    2022

  43. [43]

    Zhou, Y.; Yang, Y.; Ying, Q.; Qian, Z.; and Zhang, X. 2023. Multi-Modal Fake News Detection on Social Media via Multi-Grained Information Fusion. In Proceedings of the 2023 ACM International Conference on Multimedia Retrieval, 343--352

    2023

  44. [44]

    Zhou, Z.; Zhang, X.; Zhang, L.; Liu, J.; Cambria, E.; and Li, C. 2026. FineFake : A Knowledge-Enriched Dataset for Fine-Grained Multi-Domain Fake News Detection. Information Fusion, 132: 104253

    2026

  45. [45]

    Zhu, Y.; Sheng, Q.; Cao, J.; Li, S.; Wang, D.; and Zhuang, F. 2022. Generalizing to the Future: Mitigating Entity Bias in Fake News Detection. In Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval, 2120--2125

    2022

  46. [46]

    Zhu, Y.; Sheng, Q.; Cao, J.; Nan, Q.; Shu, K.; Wu, M.; Wang, J.; and Zhuang, F. 2023. Memory-Guided Multi-View Multi-Domain Fake News Detection. IEEE Transactions on Knowledge and Data Engineering, 35(7): 7178--7191

    2023

This paper was first reviewed by deepseek-v4-flash on August 1, 2026.