Pith. sign in

REVIEW 2 major objections 8 minor 66 references

Multi-QuAD: Multi-Level Quality-Adaptive Dynamic Network for Reliable Multimodal Classification

T0 review · 2 major / 8 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read Multi-QuAD claims that a multimodal classifier that adapts its depth and parameters to each sample's estimated quality outperforms static reliable classifiers by up to 5.9 percentage points and degrades less under Gaussian noise.

desk verdict Promising integration of prototype-based confidence with dynamic depth and parameters, but the test-time computation of T is unspecified and the missing supplement makes the large reported gains unverifiable. read the letter →

arxiv 2412.14489 v3 pith:Y3GZKXY7 submitted 2024-12-19 cs.CV

Shu Shen , C.L.Philip Chen , Tong Zhang This is my paper · ORCID
classification cs.CV
keywords multimodalclassificationreliabledynamicneuralnetworksqualityestimationnoise-freeprototypessample-adaptivedepthparametersGaussiannoiserobustness
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

Multi-QuAD argues that reliable multimodal classification should not use a fixed network for every sample. It estimates the quality of each modality and each feature of every input, then adjusts the network's depth and parameters accordingly. On four benchmarks, the paper reports that this dynamic strategy beats state-of-the-art reliable multimodal classifiers by 1.2 to 5.9 percentage points in accuracy under clean conditions, and loses substantially less accuracy under Gaussian noise. The reason to care is that reliability in safety-critical multimodal settings is usually pursued with static architectures and confidence scores; this work makes depth and parameters themselves the adaptive resource.

What carries the argument

Three components carry the argument. Noise-free Prototype Confidence Estimation (NFCE) constructs class prototypes optimized to remove noise by maximizing cross-modal mutual information and inter-class orthogonality, then produces modality-level quality $\delta_v(x_i^m)$ and feature-level quality $\delta_f(x_i^m)$ without a classifier. Global Confidence Normalized Depth (GCND) ranks $1-\delta_v(x_i^m)$ for all modalities of all samples and scales them into integer depths $D_i^m \in [1,K]$, so lower-quality modalities get deeper unimodal networks while extreme samples are reined in by the global normalization. Layer-wise Greedy Parameter (LGP) uses an LSTM unit to remember previously enhanced features and predicts the weights of all unimodal blocks at depth $t$ jointly, maximizing the feature-quality gain $\Delta_i^t$ and a sparsity loss. The central object is the quality-to-architecture mapping: per-sample quality estimates drive both depth and parameters.

What would settle it

Reproduce the NFCE and GCND depth assignments on a held-out test split of FOOD101 with labels withheld; if computing $\delta_v$ and $\delta_f$ via Eqs. (6)-(7) requires the ground-truth class, the reported 94.5% ACC and the Fig. 4 robustness curves are not attainable at test time.

Watch

Extended reading notes

Core claim

The central discovery is that a multimodal classifier can be made more reliable by making its architecture depend on a classifier-free, prototype-based estimate of input quality. Multi-QuAD computes modality-level and feature-level qualities by comparing each input with pre-optimized 'noise-free prototypes' (class centers with noise removed), avoiding the overfitting to training noise that the paper attributes to classifier-based confidence measures. The Global Confidence Normalized Depth mechanism maps these quality estimates to per-sample network depths, normalized across modalities and samples so that extreme low-quality inputs do not force excessive depth. The Layer-wise Greedy Parameter mechanism then predicts per-sample weights at each depth, layer by layer, to maximize feature-level quality gain. On BRCA, ROSMAP, CUB, and FOOD101 the paper reports ACC gains of 4.8%, 5.9%, 4.4%, and 1.2% over the strongest baseline, with smaller accuracy drops under noise.

Load-bearing premise

The test-time quality scores depend on a 'true class probability' transform that the paper does not specify for inference; if that transform requires the sample's ground-truth label, the reported accuracies would not transfer to unlabeled test data.

Editorial extensions

If this is right

  • Each modality of each sample should get its own network depth, with lower-quality inputs processed by deeper unimodal networks.
  • Network parameters should be predicted per sample and per layer, not fixed after training, so that a single block can map different feature-quality patterns to cleaner representations.
  • Confidence estimates built from noise-free prototypes rather than from the classifier maintain stable behavior across training noise levels at both modality and feature levels.
  • Dynamic multimodal networks such as DynMM gain reliability when their depth assignment is driven by normalized quality estimates rather than by intermediate gating or classifier signals.

Reading between the lines

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

  • The same prototype-based quality estimate could be reused as a drop-in confidence module for other multimodal architectures, since NFCE is trained separately from the classifier.
  • A label-free proxy for the true-class-probability transform $T$, such as maximum predicted probability or prediction entropy, would make the method deployable at test time; whether the reported margins survive that substitution is a concrete follow-up test.
  • GCND's global normalization across the whole dataset implies that reliability depends on the population distribution of input quality, not only on each sample in isolation; that principle could transfer to any per-sample adaptive system.
  • The paper's own future-work list suggests extending the same quality-adaptive mechanism from classification to regression and segmentation, where quality estimation for continuous targets would need a new definition of prototype.
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.

Referee Report

2 major / 8 minor

Summary. The paper proposes Multi-QuAD, a multimodal classification framework that estimates sample quality at modality and feature levels using noise-free prototypes (NFCE), then adapts network depth via Global Confidence Normalized Depth (GCND) and per-sample network parameters via a Layer-wise Greedy Parameter (LGP) mechanism implemented with an LSTM. Experiments on BRCA, ROSMAP, CUB, and FOOD101 report consistent accuracy gains of 4.4–5.9 points over the best static reliable-classification baselines and over the dynamic DynMM baseline, with smaller accuracy degradation under Gaussian noise. The central claim is that prototype-based quality estimates enable dynamic depth and parameter assignment that is both more accurate and more robust than static networks.

Significance. If the reported results are reproducible under a label-free inference protocol, the paper makes a strong empirical contribution to reliable multimodal classification: it combines quality estimation with dynamic architecture design, provides ablations for each component, and demonstrates consistent gains across four benchmarks, including both biomedical and vision-language data. The comparison against DynMM and the static state-of-the-art methods is useful, and the visualizations of learned biomarkers add interpretive value. The significance is therefore primarily empirical, and it hinges on the clarity and validity of the quality estimator at test time; the lack of code or the referenced supplement limits independent verification.

major comments (2)
  1. [Section III-B2, Eqs. (6)–(7)] The paper defines δ_v and δ_f using T, the “true class probability” from [4], and defers the implementation to the Supplemental Materials, which is not available with the arXiv v3 submission. These quantities are load-bearing: Eq. (8) uses δ_v to set all sample depths, and Eq. (10) plus Algorithm 1 line 17 use δ_f to train and operate LGP. The manuscript never states whether T is computed with ground-truth labels or with a label-free proxy at inference. If labels are used, the test-set results in Table I and Fig. 4 are invalid because the network is given the answer during depth and parameter assignment; if a proxy such as maximum predicted probability is used, that substitution is not disclosed or validated. Please specify the exact computation of T, state the inference-time protocol, and confirm that no test labels are used in any step of dynamic depth or parameter prediction.
  2. [Section III-B1 and III-E, Eqs. (3), (10), (13)] The prototypes ω used by NFCE are optimized on the training set, and the LGP loss maximizes Δ, which is measured as improvement in δ_f evaluated against those same prototypes. This makes the “quality gain” objective self-referential: it rewards moving features toward the training prototypes rather than toward an independent measure of held-out quality. The held-out classification results in Table II are not invalidated by this, but the interpretation that LGP “maximizes feature-level quality enhancement” would be much stronger with a validation of δ_f against held-out label correctness (e.g., correlation between δ_f and per-feature classification utility) and with reported test-set values of Δ. I frame this as a correctness-risk concern with a concrete test, not as a claim of leakage.
minor comments (8)
  1. [Section III-B2, Eq. (5)] Please specify over which axis the softmax is applied for p_f and how T reduces the class dimension to produce a per-feature δ_f; as written, p_f ∈ R^{d_m×C} and δ_f ∈ R^{d_m} leave this implicit.
  2. [Algorithm 1, line 16] The notation \hat{I}^{m+1}_i is confusing because \hat{I} indexes layer t, not modality; it should be something like \hat{I}^{m,t+1}_i to avoid implying that the (m+1)-th modality is used.
  3. [Section III-D, Eq. (10)] The phrase “such that Δ^t_i is maximized” describes a constraint, but LGPt is a learned LSTM and does not solve an optimization at inference; recommend rewording to “trained to maximize” or “aims to maximize.”
  4. [Section IV-D1 and Fig. 6] State explicitly whether the confidence evaluation on the “same clean testing samples” uses ground-truth labels; this is the same ambiguity as the T definition and should be resolved in the main text rather than only in the supplement.
  5. [Section III-C, Eq. (8)] Clarify whether the depth normalization N_K is computed once on the training set or recomputed for each test batch; this matters for deploying the model on individual samples in a streaming setting.
  6. [Figure 5, ROSMAP panel] Multi-QuAD's accuracy changes from 92.2 at K=3 to 91.6 at K=4, which looks larger than the “relatively insensitive” claim; please add a brief discussion of the variance across K.
  7. [Figures 5 and 9] The dataset name “UMPC FOOD101” appears in the figure labels, while the text and Table I use “UPMC FOOD101”; please unify the naming.
  8. [Section IV-H, Table VII] The accuracy drop is reported in parentheses for the baselines but not for Multi-QuAD's ACC*; include the same notation for Multi-QuAD to make the comparison symmetric.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the headline accuracy and robustness claims are evaluated on held-out external benchmarks, and the self-referential quality-gain objective is not the evidence for the central claim.

full rationale

The paper's central claim is empirical: Table I and Fig. 4 report held-out classification accuracy of Multi-QuAD against seven external SOTA baselines and under injected Gaussian noise. These metrics are not defined in terms of the paper's quality estimates, so the main result cannot reduce to a fitted input or to the model's own quality definition. The dynamic depth (GCND, Eq. 8) and dynamic parameters (LGP, Eq. 10) are driven by the NFCE quality estimates, but their contribution is validated by external accuracy ablations ('W/O GCND' in Fig. 5, 'W/O LGP' in Table II, and depth +/-1 in Table IV), not by the self-defined quality gain. The fact that LGP is trained to maximize the NFCE-defined feature-quality gain is a self-referential training objective, but the paper does not use that gain as evidence that LGP helps; it uses held-out accuracy. The only load-bearing caveat is that Eqs. (6)-(7) define delta_v and delta_f through T, called the true class probability, with implementation deferred to the Supplemental Materials; if T requires ground-truth labels, that would be test-time label leakage, but that is a correctness and verifiability concern, not circularity, because the reported accuracy is an external metric rather than a re-expression of T. No self-citation chain or uniqueness theorem is load-bearing, and no claim is presented as a derivation that is equivalent to its own inputs by construction.

Assumptions & free parameters 2 free parameters · 4 assumptions · 0 invented entities

The framework's headline results depend on a handful of tuned hyperparameters (notably K) and on assumptions that the prototype optimization removes noise, that the trained LSTM approximates a greedy optimum, and that quality can be computed at inference. The most serious is the unstated test-time treatment of T, which affects every reported accuracy.

free parameters (2)
  • K (max network depth) = 3
    Depth normalization range in GCND, chosen by validation on each dataset; Fig. 5 shows accuracy peaks at K=3 and the method is otherwise roughly stable for K>=3.
  • LSTM hidden state dimension hdim = Not reported
    LSTM unit in LGP; no value or selection procedure is given in the manuscript, so it is an undisclosed free parameter affecting parameter prediction capacity.
assumptions (4)
  • domain assumption Maximizing Lrob (Eq. 3) removes noise from prototypes, making them noise-free and class-discriminative.
    Stated in Sec. III-B1; only tested on synthetic Gaussian noise by tracking PN(omega), not on real-world noise or label noise.
  • ad hoc to paper The LSTM-based LGPt trained with the combined loss actually solves the constrained greedy optimization in Eq. (10), i.e., it maximizes the per-layer quality gain.
    Sec. III-D presents Eq. (10) as a constraint to be maximized, but Algorithm 1 trains an LSTM with an auxiliary loss; no optimality or regret bound is given, so the greedy optimality is assumed.
  • ad hoc to paper The true-class probability T can be evaluated at test time without ground-truth labels.
    Sec. III-B2 Eq. (6)-(7); the implementation is deferred to the Supplement, leaving a label-free inference protocol unstated.
  • domain assumption Element-wise gating q * I preserves informative features and removes redundant ones, so deeper blocks monotonically improve feature quality.
    Sec. III-A Eq. (2), inherited from prior dynamic gating networks; not independently validated here beyond t-SNE visualizations.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Multi-QuAD: Multi-Level Quality-Adaptive Dynamic Network for Reliable Multimodal Classification." pith.science (2026). https://pith.science/paper/Y3GZKXY7

@misc{pith2026241214489,
  author       = {Pith},
  title        = {Pith review of: Multi-QuAD: Multi-Level Quality-Adaptive Dynamic Network for Reliable Multimodal Classification},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/Y3GZKXY7}},
  note         = {Machine review of arXiv:2412.14489}
}
read the original abstract

Multimodal machine learning has achieved remarkable progress in many scenarios, but its reliability is undermined by varying sample quality. This paper finds that existing reliable multimodal classification methods not only fail to provide robust estimation of data quality, but also lack dynamic networks for sample-specific depth and parameters to achieve reliable inference. To this end, a novel framework for multimodal reliable classification termed \textit{Multi-level Quality-Adaptive Dynamic multimodal network} (Multi-QuAD) is proposed. Multi-QuAD first adopts a novel approach based on noise-free prototypes and a classifier-free design to reliably estimate the quality of each sample at both modality and feature levels. It then achieves sample-specific network depth via the \textbf{\textit{Global Confidence Normalized Depth (GCND)}} mechanism. By normalizing depth across modalities and samples, \textit{\textbf{GCND}} effectively mitigates the impact of challenging modality inputs on dynamic depth reliability. Furthermore, Multi-QuAD provides sample-adaptive network parameters via the \textbf{\textit{Layer-wise Greedy Parameter (LGP)}} mechanism driven by feature-level quality. The cross-modality layer-wise greedy strategy in \textbf{\textit{LGP}} designs a reliable parameter prediction paradigm for multimodal networks with variable architecture for the first time. Experiments conducted on four datasets demonstrate that Multi-QuAD significantly outperforms state-of-the-art methods in classification performance and reliability, exhibiting strong adaptability to data with diverse quality.

Figures

Figures reproduced from arXiv: 2412.14489 by the authors.

Figure 1
Figure 1. Empirical studies under varying data quality. We simulate data quality [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. The framework of the proposed Multi-QuAD (better viewed in colour). Without loss of generality, this figure illustrates the case of two modalities, with [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. The detailed implementation of cross-modality greedy parameter [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (7 more)
Figure 4
Figure 4. Figure 4: Comparison of classification accuracy of different models under [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]
Figure 5
Figure 5. Figure 5: Ablation study to demonstrate the effectiveness of [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]
Figure 6
Figure 6. Figure 6: Comparison of the sensitivity of some commonly used traditional [PITH_FULL_IMAGE:figures/full_fig_p007_6.png]
Figure 7
Figure 7. Figure 7: The probability of noise-free prototypes in the noise distribution [PITH_FULL_IMAGE:figures/full_fig_p008_7.png]
Figure 8
Figure 8. Figure 8: Results of the modality-level and feature-level quality estimated by [PITH_FULL_IMAGE:figures/full_fig_p009_8.png]
Figure 9
Figure 9. Figure 9: Visualization of the parameter values of different samples on UMPC [PITH_FULL_IMAGE:figures/full_fig_p009_9.png]
Figure 11
Figure 11. Figure 11: The t-SNE visualization results of the output feature vectors of [PITH_FULL_IMAGE:figures/full_fig_p010_11.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

66 extracted references · 49 canonical work pages

  1. [4]

    Multimodal dynamics: Dynamical fusion for trustworthy multimodal classification,

    Z. Han, F. Yang, J. Huang, C. Zhang, and J. Yao, “Multimodal dynamics: Dynamical fusion for trustworthy multimodal classification,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2022, pp. 20 707–20 717

    work page 2022

  2. [1]

    Zheng, M

    X. Zheng, M. Wang, K. Huang, and E. Zhu, “Global and cross-modal feature aggregation for multi-omics data classification and application JOURNAL OF LATEX CLASS FILES, VOL. 14, NO. 8, MAY 2025 11 on drug response prediction,” Information Fusion, vol. 102, p. 102077, 2024

    work page 2025

  3. [2]

    Predictive dynamic fusion,

    B. Cao, Y . Xia, Y . Ding, C. Zhang, and Q. Hu, “Predictive dynamic fusion,” in Proceedings of the 41st International Conference on Machine Learning , ser. Proceedings of Machine Learning Research, R. Salakhutdinov, Z. Kolter, K. Heller, A. Weller, N. Oliver, J. Scarlett, and F. Berkenkamp, Eds., vol. 235. PMLR, 21–27 Jul 2024, pp. 5608–5628. [Online]. Av...

    work page 2024

  4. [3]

    Calm: An enhanced encoding and confidence evaluating framework for trustworthy multi-view learning,

    H. Zhou, Z. Xue, Y . Liu, B. Li, J. Du, M. Liang, and Y . Qi, “Calm: An enhanced encoding and confidence evaluating framework for trustworthy multi-view learning,” in Proceedings of the 31st ACM International Conference on Multimedia , 2023, pp. 3108–3116

    work page 2023

  5. [5]

    An empirical study of training end-to-end vision-and-language transformers,

    Z.-Y . Dou, Y . Xu, Z. Gan, J. Wang, S. Wang, L. Wang, C. Zhu, P. Zhang, L. Yuan, N. Peng et al. , “An empirical study of training end-to-end vision-and-language transformers,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2022, pp. 18 166–18 176

    work page 2022

  6. [6]

    Clip-event: Connecting text and images with event structures,

    M. Li, R. Xu, S. Wang, L. Zhou, X. Lin, C. Zhu, M. Zeng, H. Ji, and S.-F. Chang, “Clip-event: Connecting text and images with event structures,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , 2022, pp. 16 420–16 429

    work page 2022

  7. [7]

    Cubemlp: An mlp-based model for multimodal sentiment analysis and depression estimation,

    H. Sun, H. Wang, J. Liu, Y .-W. Chen, and L. Lin, “Cubemlp: An mlp-based model for multimodal sentiment analysis and depression estimation,” in Proceedings of the 30th ACM international conference on multimedia, 2022, pp. 3722–3729

    work page 2022

  8. [8]

    Clap learning audio concepts from natural language supervision,

    B. Elizalde, S. Deshmukh, M. Al Ismail, and H. Wang, “Clap learning audio concepts from natural language supervision,” in ICASSP 2023- 2023 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). IEEE, 2023, pp. 1–5

    2023

Show all 66 references
  1. [9]

    Unimf: A unified multimodal framework for multimodal sentiment analysis in missing modalities and unaligned multimodal sequences,

    R. Huan, G. Zhong, P. Chen, and R. Liang, “Unimf: A unified multimodal framework for multimodal sentiment analysis in missing modalities and unaligned multimodal sequences,” IEEE Transactions on Multimedia, vol. 26, pp. 5753–5768, 2024

    2024

  2. [10]

    Building multimodal knowledge bases with multimodal computational sequences and generative adversarial net- works,

    D. Chen and R. Zhang, “Building multimodal knowledge bases with multimodal computational sequences and generative adversarial net- works,” IEEE Transactions on Multimedia , vol. 26, pp. 2027–2040, 2024

    2027

  3. [11]

    Deep multimodal sequence fusion by regularized expressive representation distillation,

    X. Guo, A. W.-K. Kong, and A. Kot, “Deep multimodal sequence fusion by regularized expressive representation distillation,” IEEE Transactions on Multimedia, vol. 25, pp. 2085–2096, 2023

    2023

  4. [12]

    Msdlf-k: A multimodal feature learning approach for sentiment analysis in korean incorporating text and speech,

    T.-Y . Kim, J. Yang, and E. Park, “Msdlf-k: A multimodal feature learning approach for sentiment analysis in korean incorporating text and speech,” IEEE Transactions on Multimedia , vol. 27, pp. 1266–1276, 2025

    2025

  5. [13]

    Sia-net: Sparse interactive attention network for multimodal emotion recognition,

    S. Li, T. Zhang, and C. L. P. Chen, “Sia-net: Sparse interactive attention network for multimodal emotion recognition,” IEEE Transactions on Computational Social Systems , vol. 11, no. 5, pp. 6782–6794, 2024

    2024

  6. [14]

    Aia-net: Adaptive interactive attention network for text–audio emotion recogni- tion,

    T. Zhang, S. Li, B. Chen, H. Yuan, and C. L. Philip Chen, “Aia-net: Adaptive interactive attention network for text–audio emotion recogni- tion,” IEEE Transactions on Cybernetics, vol. 53, no. 12, pp. 7659–7671, 2023

    2023

  7. [15]

    Mia-net: Multi-modal interactive attention network for multi-modal affective analysis,

    S. Li, T. Zhang, B. Chen, and C. L. P. Chen, “Mia-net: Multi-modal interactive attention network for multi-modal affective analysis,” IEEE Transactions on Affective Computing , vol. 14, no. 4, pp. 2796–2809, 2023

    2023

  8. [16]

    Robust saliency- aware distillation for few-shot fine-grained visual recognition,

    H. Liu, C. L. P. Chen, X. Gong, and T. Zhang, “Robust saliency- aware distillation for few-shot fine-grained visual recognition,” IEEE Transactions on Multimedia , vol. 26, pp. 7529–7542, 2024

    2024

  9. [17]

    Videodreamer: Customized multi-subject text-to-video generation with disen-mix finetuning on language-video foundation models,

    H. Chen, X. Wang, G. Zeng, Y . Zhang, Y . Zhou, F. Han, Y . Wu, and W. Zhu, “Videodreamer: Customized multi-subject text-to-video generation with disen-mix finetuning on language-video foundation models,” IEEE Transactions on Multimedia , pp. 1–11, 2025

    2025

  10. [18]

    Dual-task mutual reinforcing embedded joint video paragraph retrieval and grounding,

    M. Wang, H. Li, Y . Zhang, J. Li, M. Xie, and D. Tao, “Dual-task mutual reinforcing embedded joint video paragraph retrieval and grounding,” IEEE Transactions on Multimedia , pp. 1–16, 2025

    2025

  11. [19]

    Motion direction awareness: A biomimetic dynamic capture mechanism for video prediction,

    L. Gan, J. Lai, J. Zhu, H. Liu, and L. Gao, “Motion direction awareness: A biomimetic dynamic capture mechanism for video prediction,” IEEE Transactions on Multimedia , pp. 1–15, 2025

    2025

  12. [20]

    Balanced multimodal learning via on-the-fly gradient modulation,

    X. Peng, Y . Wei, A. Deng, D. Wang, and D. Hu, “Balanced multimodal learning via on-the-fly gradient modulation,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2022, pp. 8238–8247

    2022

  13. [21]

    Cal- ibrating multimodal learning,

    H. Ma, Q. Zhang, C. Zhang, B. Wu, H. Fu, J. T. Zhou, and Q. Hu, “Cal- ibrating multimodal learning,” in International Conference on Machine Learning. PMLR, 2023, pp. 23 429–23 450

    2023

  14. [22]

    On-the-fly modulation for balanced multimodal learning,

    Y . Wei, D. Hu, H. Du, and J.-R. Wen, “On-the-fly modulation for balanced multimodal learning,” IEEE Transactions on Pattern Analysis and Machine Intelligence , 2024

    2024

  15. [23]

    Noise-aware unsupervised deep lidar-stereo fusion,

    X. Cheng, Y . Zhong, Y . Dai, P. Ji, and H. Li, “Noise-aware unsupervised deep lidar-stereo fusion,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2019, pp. 6339–6348

    2019

  16. [24]

    Uncertainty-aware multi-view representation learning,

    Y . Geng, Z. Han, C. Zhang, and Q. Hu, “Uncertainty-aware multi-view representation learning,” in Proceedings of the AAAI Conference on Artificial Intelligence, vol. 35, no. 9, 2021, pp. 7545–7553

    2021

  17. [25]

    Multimodal fusion on low-quality data: A comprehensive survey,

    Q. Zhang, Y . Wei, Z. Han, H. Fu, X. Peng, C. Deng, Q. Hu, C. Xu, J. Wen, D. Hu et al. , “Multimodal fusion on low-quality data: A comprehensive survey,” arXiv preprint arXiv:2404.18947 , 2024

    2024 arXiv

  18. [26]

    Towards safe autonomous driving: Capture uncertainty in the deep neural network for lidar 3d vehicle detection,

    D. Feng, L. Rosenbaum, and K. Dietmayer, “Towards safe autonomous driving: Capture uncertainty in the deep neural network for lidar 3d vehicle detection,” in 2018 21st international conference on intelligent transportation systems (ITSC) . IEEE, 2018, pp. 3266–3273

    2018

  19. [27]

    Exploring uncertainty measures in deep networks for multiple sclerosis lesion detection and segmentation,

    T. Nair, D. Precup, D. L. Arnold, and T. Arbel, “Exploring uncertainty measures in deep networks for multiple sclerosis lesion detection and segmentation,” Medical image analysis , vol. 59, p. 101557, 2020

    2020

  20. [28]

    Dpnet: Dynamic poly-attention network for trustworthy multi-modal classification,

    X. Zou, C. Tang, X. Zheng, Z. Li, X. He, S. An, and X. Liu, “Dpnet: Dynamic poly-attention network for trustworthy multi-modal classification,” in Proceedings of the 31st ACM International Conference on Multimedia, 2023, pp. 3550–3559

    2023

  21. [29]

    Multi-level confi- dence learning for trustworthy multimodal classification,

    X. Zheng, C. Tang, Z. Wan, C. Hu, and W. Zhang, “Multi-level confi- dence learning for trustworthy multimodal classification,” inProceedings of the AAAI Conference on Artificial Intelligence , vol. 37, no. 9, 2023, pp. 11 381–11 389

    2023

  22. [30]

    Cold fusion: Calibrated and ordinal latent dis- tribution fusion for uncertainty-aware multimodal emotion recognition,

    M. K. Tellamekala, S. Amiriparian, B. W. Schuller, E. Andr ´e, T. Gies- brecht, and M. Valstar, “Cold fusion: Calibrated and ordinal latent dis- tribution fusion for uncertainty-aware multimodal emotion recognition,” IEEE Transactions on Pattern Analysis and Machine Intelligen...

    2023

  23. [31]

    Provable dynamic fusion for low-quality multimodal data,

    Q. Zhang, H. Wu, C. Zhang, Q. Hu, H. Fu, J. T. Zhou, and X. Peng, “Provable dynamic fusion for low-quality multimodal data,” in Inter- national conference on machine learning . PMLR, 2023, pp. 41 753– 41 769

    2023

  24. [32]

    Pace-adaptive and noise-resistant contrastive learning for multimodal feature fusion,

    X. Guo, A. Kot, and A. W.-K. Kong, “Pace-adaptive and noise-resistant contrastive learning for multimodal feature fusion,” IEEE Transactions on Multimedia, vol. 25, pp. 9437–9448, 2023

    2023

  25. [33]

    Disentangled multimodal representation learning for recommendation,

    F. Liu, H. Chen, Z. Cheng, A. Liu, L. Nie, and M. Kankanhalli, “Disentangled multimodal representation learning for recommendation,” IEEE Transactions on Multimedia , vol. 25, pp. 7149–7159, 2023

    2023

  26. [34]

    Multimodal affective computing with dense fusion transformer for inter- and intra-modality interactions,

    H. Deng, Z. Yang, T. Hao, Q. Li, and W. Liu, “Multimodal affective computing with dense fusion transformer for inter- and intra-modality interactions,” IEEE Transactions on Multimedia, vol. 25, pp. 6575–6587, 2023

    2023

  27. [35]

    Ld-man: Layout-driven multimodal attention network for online news sentiment recognition,

    W. Guo, Y . Zhang, X. Cai, L. Meng, J. Yang, and X. Yuan, “Ld-man: Layout-driven multimodal attention network for online news sentiment recognition,” IEEE Transactions on Multimedia, vol. 23, pp. 1785–1798, 2021

    2021

  28. [36]

    Image-text multimodal emotion classification via multi-view attentional network,

    X. Yang, S. Feng, D. Wang, and Y . Zhang, “Image-text multimodal emotion classification via multi-view attentional network,” IEEE Trans- actions on Multimedia , vol. 23, pp. 4014–4026, 2021

    2021

  29. [37]

    Congmc: Consistency-guided multi- modal clustering via mutual information maximin,

    Y . Mao, X. Yan, J. Liu, and Y . Ye, “Congmc: Consistency-guided multi- modal clustering via mutual information maximin,” IEEE Transactions on Multimedia, vol. 26, pp. 5131–5146, 2024

    2024

  30. [38]

    Dynamically shifting multimodal representations via hybrid-modal attention for multimodal sentiment analysis,

    R. Lin and H. Hu, “Dynamically shifting multimodal representations via hybrid-modal attention for multimodal sentiment analysis,” IEEE Transactions on Multimedia , vol. 26, pp. 2740–2755, 2024

    2024

  31. [39]

    Dominant single-modal supplementary fusion (simsuf) for multimodal sentiment analysis,

    J. Huang, Y . Ji, Z. Qin, Y . Yang, and H. T. Shen, “Dominant single-modal supplementary fusion (simsuf) for multimodal sentiment analysis,” IEEE Transactions on Multimedia , vol. 26, pp. 8383–8394, 2024

    2024

  32. [40]

    Trusted multi-view classifica- tion,

    Z. Han, C. Zhang, H. Fu, and J. T. Zhou, “Trusted multi-view classifica- tion,” in International Conference on Learning Representations , 2020

    2020

  33. [41]

    Transforming neural-net output levels to probability distributions,

    J. Denker and Y . LeCun, “Transforming neural-net output levels to probability distributions,” Advances in neural information processing systems, vol. 3, 1990

    1990

  34. [42]

    D. J. C. Mackay, Bayesian methods for adaptive models . California Institute of Technology, 1992

    1992

  35. [43]

    R. M. Neal, Bayesian learning for neural networks . Springer Science & Business Media, 2012, vol. 118

    2012

  36. [44]

    A baseline for detecting misclassified and out-of-distribution examples in neural networks,

    D. Hendrycks and K. Gimpel, “A baseline for detecting misclassified and out-of-distribution examples in neural networks,” arXiv preprint arXiv:1610.02136, 2016. JOURNAL OF LATEX CLASS FILES, VOL. 14, NO. 8, MAY 2025 12

    2016 arXiv

  37. [45]

    Energy-based out-of-distribution detection,

    W. Liu, X. Wang, J. Owens, and Y . Li, “Energy-based out-of-distribution detection,” Advances in neural information processing systems , vol. 33, pp. 21 464–21 475, 2020

    2020

  38. [46]

    Convolutional networks with adaptive infer- ence graphs,

    A. Veit and S. Belongie, “Convolutional networks with adaptive infer- ence graphs,” in Proceedings of the European conference on computer vision (ECCV), 2018, pp. 3–18

    2018

  39. [47]

    Channel gating neural networks,

    W. Hua, Y . Zhou, C. M. De Sa, Z. Zhang, and G. E. Suh, “Channel gating neural networks,” Advances in Neural Information Processing Systems, vol. 32, 2019

    2019

  40. [48]

    Pixel-adaptive convolutional neural networks,

    H. Su, V . Jampani, D. Sun, O. Gallo, E. Learned-Miller, and J. Kautz, “Pixel-adaptive convolutional neural networks,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2019, pp. 11 166–11 175

    2019

  41. [49]

    Dynamic neural networks: A survey,

    Y . Han, G. Huang, S. Song, L. Yang, H. Wang, and Y . Wang, “Dynamic neural networks: A survey,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 44, no. 11, pp. 7436–7456, 2021

    2021

  42. [50]

    Multi-scale dense networks for resource efficient image classification,

    G. Huang, D. Chen, T. Li, F. Wu, L. Van Der Maaten, and K. Q. Weinberger, “Multi-scale dense networks for resource efficient image classification,” arXiv preprint arXiv:1703.09844 , 2017

    2017 arXiv

  43. [51]

    Skipnet: Learning dynamic routing in convolutional networks,

    X. Wang, F. Yu, Z.-Y . Dou, T. Darrell, and J. E. Gonzalez, “Skipnet: Learning dynamic routing in convolutional networks,” in Proceedings of the European conference on computer vision (ECCV) , 2018, pp. 409– 424

    2018

  44. [52]

    Anytime recognition with routing convolutional networks,

    Z. Jie, P. Sun, X. Li, J. Feng, and W. Liu, “Anytime recognition with routing convolutional networks,” IEEE transactions on pattern analysis and machine intelligence , vol. 43, no. 6, pp. 1875–1886, 2019

    2019

  45. [53]

    Deebert: Dynamic early ex- iting for accelerating bert inference,

    J. Xin, R. Tang, J. Lee, Y . Yu, and J. Lin, “Deebert: Dynamic early ex- iting for accelerating bert inference,” arXiv preprint arXiv:2004.12993 , 2020

    2004 arXiv

  46. [54]

    Bert loses patience: Fast and robust inference with early exit,

    W. Zhou, C. Xu, T. Ge, J. McAuley, K. Xu, and F. Wei, “Bert loses patience: Fast and robust inference with early exit,” Advances in Neural Information Processing Systems , vol. 33, pp. 18 330–18 341, 2020

    2020

  47. [55]

    Dynamicdet: A unified dynamic architecture for object detection,

    Z. Lin, Y . Wang, J. Zhang, and X. Chu, “Dynamicdet: A unified dynamic architecture for object detection,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2023, pp. 6282–6291

    2023

  48. [56]

    Dynamic multimodal fusion,

    Z. Xue and R. Marculescu, “Dynamic multimodal fusion,” in Proceed- ings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2023, pp. 2575–2584

    2023

  49. [57]

    Learning robust representations via multi-view information bottleneck,

    M. Federici, A. Dutta, P. Forr ´e, N. Kushman, and Z. Akata, “Learning robust representations via multi-view information bottleneck,” arXiv preprint arXiv:2002.07017, 2020

    2002 arXiv

  50. [58]

    Deep semantic clustering by partition confidence maximisation,

    J. Huang, S. Gong, and X. Zhu, “Deep semantic clustering by partition confidence maximisation,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , 2020, pp. 8849–8858

    2020

  51. [59]

    Overview and findings from the religious orders study,

    D. A Bennett, J. A Schneider, Z. Arvanitakis, and R. S Wilson, “Overview and findings from the religious orders study,” Current Alzheimer Research, vol. 9, no. 6, pp. 628–645, 2012

    2012

  52. [60]

    A multi-omic atlas of the human frontal cortex for aging and alzheimer’s disease research. sci data 5: 180142,

    P. De Jager, Y . Ma, C. McCabe, J. Xu, B. Vardarajan, D. Felsky, H. Klein, C. White, M. Peters, B. Lodgson et al. , “A multi-omic atlas of the human frontal cortex for aging and alzheimer’s disease research. sci data 5: 180142,” 2018

    2018

  53. [61]

    The caltech-ucsd birds-200-2011 dataset,

    C. Wah, S. Branson, P. Welinder, P. Perona, and S. Belongie, “The caltech-ucsd birds-200-2011 dataset,” 2011

    2011

  54. [62]

    Recipe recognition with large multimodal food dataset,

    X. Wang, D. Kumar, N. Thome, M. Cord, and F. Precioso, “Recipe recognition with large multimodal food dataset,” in 2015 IEEE Interna- tional Conference on Multimedia & Expo Workshops (ICMEW) . IEEE, 2015, pp. 1–6

    2015

  55. [63]

    Trustworthy multimodal regression with mixture of normal-inverse gamma distribu- tions,

    H. Ma, Z. Han, C. Zhang, H. Fu, J. T. Zhou, and Q. Hu, “Trustworthy multimodal regression with mixture of normal-inverse gamma distribu- tions,” Advances in Neural Information Processing Systems , vol. 34, pp. 6881–6893, 2021

    2021

  56. [64]

    Epigenetic-mediated downregulation of zinc finger protein 671 (znf671) predicts poor prognosis in multiple solid tumors,

    J. Zhang, Z. Zheng, J. Zheng, T. Xie, Y . Tian, R. Li, B. Wang, J. Lin, A. Xu, X. Huang et al. , “Epigenetic-mediated downregulation of zinc finger protein 671 (znf671) predicts poor prognosis in multiple solid tumors,” Frontiers in Oncology, vol. 9, p. 342, 2019

    2019

  57. [65]

    mirna-378 is downregulated by xbp1 and inhibits growth and migration of luminal breast cancer cells,

    V . Arabkari, D. Barua, M. M. Hossain, M. Webber, T. Smith, A. Gupta, and S. Gupta, “mirna-378 is downregulated by xbp1 and inhibits growth and migration of luminal breast cancer cells,” International Journal of Molecular Sciences, vol. 25, no. 1, p. 186, 2023

    2023

  58. [66]

    Clinical relevance of the deregulated kallikrein-related peptidase 8 mrna expression in breast cancer: a novel independent indicator of disease-free survival,

    K. Michaelidou, A. Ardavanis, and A. Scorilas, “Clinical relevance of the deregulated kallikrein-related peptidase 8 mrna expression in breast cancer: a novel independent indicator of disease-free survival,” Breast cancer research and treatment , vol. 152, pp. 323–336, 2015

    2015

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.