Pith. sign in

REVIEW 5 major objections 8 minor 25 references

Latent label distribution grid representation for modeling uncertainty

T0 review · 5 major / 8 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read A 3-D label grid turns noisy labels into a low-noise target, the paper argues.

desk verdict A broadly evaluated but internally inconsistent grid representation for label uncertainty; the mechanism fails as written, though the empirical work is substantial. read the letter →

arxiv 2505.21180 v1 pith:MEHZZW3U submitted 2025-05-27 cs.LG cs.AI

ShuNing Sun , YinSong Xiong , Yu Zhang , Zhuoran Zheng This is my paper
classification cs.LGcs.AI
keywords labeldistributionlearninglatentrepresentationuncertaintyGaussianexpansionTuckerreconstructionlow-rankregularizationnoiserobustnessMedMNIST
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

Label distribution learning assumes annotations are clean, but real label distributions are expensive to collect and often inexact. This paper proposes a latent label distribution grid (LLDG): a three-dimensional tensor in which every pairwise label difference is expanded into a Gaussian vector, so the grid encodes both label relations and their uncertainty. The learned grid is pulled toward this Gaussian prior and simultaneously passed through a Tucker decomposition–reconstruction step to enforce low rank and strip noise. The central claim is that this grid representation forms a low-noise learning space, yielding more accurate and noise-robust label distributions than existing LDL methods on a series of benchmarks.

What carries the argument

The central object is the latent label distribution grid (LLDG): a $c\times c\times c$ tensor $B$. Its construction combines (1) a label difference matrix $D$ with entries $a_{ij}=d^j-d^i$ from the true label distribution; (2) a Gaussian expansion that turns each entry into a length-$c$ vector with mean $a_{ij}$ and variance $1-|a_{ij}|$, producing the prior grid $\hat{B}$; and (3) a Tucker decomposition–reconstruction operator $B^*=\operatorname{Tucker}^{[R]}(\operatorname{Tucker}^{[D]}(B))$ that enforces low rank on the learned grid. A local-global feature extractor (1D convolutions plus a Transformer) generates $B$ from the input features, and an LLDG-Mixer with three linear layers squeezes the grid along its axes to output the label distribution. The loss combines an L2 label-distribution error with an L1 grid-prior error, weighted by 0.5.

What would settle it

Train the same pipeline with the grid prior computed from randomly shuffled label values, so the difference matrix no longer reflects true label relations. If accuracy and noise robustness stay essentially unchanged, the specific label-difference prior is not what carries the result; if they degrade, the prior is doing real work. A second check is to sweep the Tucker ranks: if performance is flat across very different ranks, the low-rank constraint is not the active noise-reduction mechanism.

Watch

Extended reading notes

Core claim

The paper's central claim is that modeling uncertainty directly in the label representation space—rather than only in the loss function—improves LDL accuracy and robustness. To do this, it builds a $c\times c\times c$ grid $B$ from the sample's features, then regularizes $B$ toward a prior grid $\hat{B}$ obtained by taking the pairwise difference matrix $D$ of the ground-truth label distribution and replacing each entry $a_{ij}$ by a length-$c$ Gaussian sample with mean $a_{ij}$ and variance $1-|a_{ij}|$. Because the grid is a tensor, a Tucker decomposition–reconstruction scheme enforces low-rank structure, which the authors describe as noise reduction. A small mixer network then collapses the regularized grid into a predicted label distribution. The authors report the method outperforms or matches seven LDL baselines on 15 datasets, degrades gracefully under added Gaussian noise, and transfers competitively to MedMNIST classification.

Load-bearing premise

The method's success rests on the assumption that a Gaussian prior built from ground-truth label differences—with variance set to one minus the absolute difference—captures the true uncertainty structure of the label space, so that pulling the learned grid toward it removes noise rather than information.

Editorial extensions

If this is right

  • On the 15 LDL datasets, the method posts the best or tied-best score on the majority of the six evaluation metrics, with the largest margins on wc-LDL and SBU-3DFE.
  • The grid representation keeps the model stable under Gaussian label noise with standard deviation up to 1.0, with only small metric degradation.
  • LLDG can serve as an upstream representation for classification: on MedMNIST it achieves the best or near-best AUC and accuracy on most of the ten tasks among the reported methods.
  • Ablation on the Human Gene dataset shows that removing the grid or the Tucker reconstruction worsens all metrics, supporting that both components contribute.
  • Because the grid is built from labels alone, the representation can be attached to any feature extractor, not just the proposed CFormer.

Reading between the lines

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

  • Inference: the Gaussian prior is anchored to ground-truth label differences, which makes the training signal depend on label quality; a natural extension is to bootstrap the prior from predicted distributions for unlabeled or partially labeled samples.
  • Inference: the variance rule $1-|a_{ij}|$ asserts that similar labels are confidently related and dissimilar labels are uncertain; testing alternative variance schedules (for instance, variance proportional to observed disagreement) could show whether this specific schedule matters.
  • Inference: the reported gains are small in absolute terms on several Yeast datasets, so a practical reading is that LLDG buys robustness more than raw accuracy; comparing area under the curve over noisy regimes rather than point metrics would clarify the trade-off.
  • Inference: since the Tucker ranks are never specified, the method's reproducibility depends on them; an explicit rank-selection rule would turn the method into a drop-in module for other LDL pipelines.
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

5 major / 8 minor

Summary. The paper proposes a Latent Label Distribution Grid (LLDG) representation for label distribution learning (LDL). For each sample, a pairwise label difference matrix D is computed from the ground-truth label distribution, and each entry a_ij is expanded into a length-c vector sampled from a Gaussian with mean a_ij and variance 1-|a_ij|, forming a c×c×c target grid B_hat. A network with 1D convolutions and a Transformer produces a grid B bounded by tanh to (-1,1); B is regularized by B_hat through an L2 loss and by a low-rank Tucker reconstruction. The regularized grid is then compressed by an LLDG-Mixer into a predicted label distribution. Experiments cover 15 LDL datasets and a MedMNIST classification extension, with claims of competitive accuracy, noise robustness, and stable representation.

Significance. If the construction were fully validated, the combination of label-relation expansion, Gaussian uncertainty modeling, and Tucker low-rank regularization would be a novel representation-level contribution to LDL. The paper has several concrete strengths: it compares against seven LDL baselines on 15 datasets, includes an ablation study, a noise-disturbance experiment, an extension to MedMNIST, and states that code and datasets are released. The low-rank tensor regularization is a distinctive ingredient. However, the current empirical support is weakened by an apparent data-entry error, statistically insignificant ablation differences, and an underexplained classification extension; the central representational mechanism also has a mismatch between the unbounded Gaussian target and the bounded generated grid. These issues are fixable but are load-bearing for the paper's claims.

major comments (5)
  1. [§3, Eq. (1), Eq. (2), Eq. (8)] The generated grid B is constrained by tanh to lie in (-1,1), while the target grid B_hat is sampled from a Gaussian with mean a_ij and variance 1-|a_ij|. For diagonal entries a_ij=0, the variance is 1, so about 32% of sampled values lie outside (-1,1); for off-diagonal entries with |a_ij| between roughly 0.5 and 0.9, the fraction outside (-1,1) ranges from about 26% to 38%. No clipping or truncation of B_hat is specified. Consequently, a substantial part of the target cannot be represented by B, the L2 grid loss in Eq. (8) cannot approach zero by construction, and the learned B is at best a projection of B_hat onto [-1,1]. This contradicts the paper's description of B being 'bounded ... to avoid outliers' while being driven by the Gaussian prior. Please specify the target construction precisely (e.g., truncation, a different activation, or a principled reason for allowing unrepresentable targets), or the claimed 'low-noise representation space' is not realized as written.
  2. [Table 3, Yeast-cdc row] The reported Chebyshev distance for Ours on Yeast-cdc is 0.1587±0.0004, which is an order of magnitude worse than all baselines (which are near 0.016) and inconsistent with the paper's claim that LLDG outperforms the neural-network baselines on most datasets. The tight standard deviation and the healthy values on the other five metrics strongly suggest a typo, most likely a missing leading zero (0.01587). Please correct the entry and re-verify the corresponding comparison; as printed, this single row undermines the cross-dataset claim.
  3. [Table 5 and §4, Ablation study] The ablation results do not support the stated conclusion that 'the Tucker reconstruction algorithm has significant benefits.' For example, Chebyshev is 0.0522±0.0011 with LLDG versus 0.0524±0.0009 without LLDG and 0.0529±0.0011 without Tucker; Cosine is 0.8368±0.0027 versus 0.8346±0.0019 versus 0.8349±0.0008. These differences are all within one standard deviation. Please report significance tests, confidence intervals, or repeated-run distributions, and temper the claim accordingly. In addition, the 'w/o LLDG' condition is described only as 'modify the output dimension of LLDG-Mixer and remove the Lossg'; this does not specify what remains of the grid architecture, making the ablation difficult to interpret.
  4. [§4, Potential of LLDG and Table 7] The MedMNIST extension is a central part of the claimed generality, but the experimental setup is not reproducible from the text. The paper says 'for every single label, by modeling the Gaussian prior is expanded into a vector shape as a learning target for the LLDG,' yet it does not specify how integer class labels are converted into label distributions, what the target grid is for a classification task, what classifier head is used after the grid, or the network input representation for medical images. Without these details, Table 7 cannot be checked or reproduced. Please provide a complete description, ideally in the main text or appendix, including any label-distribution construction and training hyperparameters.
  5. [§4, Noise disturbance and Table 6] The noise-robustness claim is not supported by the reported experiment. Table 6 shows only LLDG's performance under Gaussian noise of increasing standard deviation; no baseline method is evaluated under the same noise conditions. The statement that 'it is still competitive compared to other methods' therefore has no empirical basis in the table. Please add comparable baselines under the same noise protocol, or explicitly restrict the claim to the stability of LLDG itself.
minor comments (8)
  1. [§3, Eq. (8)] Equation (8) contains a stray closing parenthesis: '||Bijk − ˆBijk )||2' should be '||Bijk − ˆBijk||2'.
  2. [§3, Eq. (3)] In the text describing Tucker decomposition, the second factor matrix is called 'U' twice, and then 'V' is introduced; the factor matrices in Eq. (3) are U, V, and X, so the text should say V for the second factor.
  3. [§4, Visualization and Table 4] In Table 4, the SBU-3DFE rows for IIS-LLD and Duo-LDL are identical to four decimal places across all metrics, which appears to be a transcription error; please verify these entries.
  4. [§4, Noise disturbance] The text says 'a Gaussian noise with a variance of 1 is added,' but Table 6 reports standard deviations of 0.1, 0.2, 0.5, and 1.0. Please clarify whether the parameter is the standard deviation or the variance.
  5. [§4, Visualization] The dataset name 'Yest-dtt' in the visualization section should be 'Yeast-dtt'.
  6. [§4, Ablation study] The sentence 'The experimental results are summarized in Table 4' should refer to Table 5, which contains the ablation results.
  7. [§3, Latent label distribution grid] The paper states 'we do not make any prior assumptions about label correlation' but then constructs a label correlation matrix based on differences between label values; this is itself a modeling assumption and should be acknowledged.
  8. [§3, Latent label distribution grid] The variance choice 1-|a_ij| is justified by saying 'the clearer the relationship between label values, the smaller the uncertainty,' but under this rule equal label values (a_ij=0) receive the largest variance, which seems to say that identical labels have the most uncertain relationship. Please clarify the intended interpretation.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity found: the label-derived Gaussian grid is a supervised auxiliary target, not a prediction-by-construction, and the self-citations are motivational rather than load-bearing.

full rationale

The derivation chain is self-contained. Eq. (2) builds the target grid B^ by Gaussian sampling of the ground-truth label difference matrix D, and Eq. (8) trains the feature-generated grid B to match it. This is an ordinary supervised auxiliary loss, not a case where a predicted quantity equals a fitted input by construction: at inference, B and the predicted label distribution depend only on features X, not on ground-truth labels. The paper's use of its own prior work ([6], [25]) motivates the Gaussian-expansion idea, but the construction is fully specified in Eqs. (1)-(9); no uniqueness theorem or external result is imported as load-bearing. The Tanh-bounded B vs. unbounded Gaussian B^ mismatch is a real representability/correctness concern (Loss_g cannot vanish and B can at best be a clipped projection of B^), but that is not a circularity: it neither makes a fitted parameter masquerade as a prediction nor makes the derivation equivalent to its input by definition. Benchmark evaluation is external to the fitted values, so the paper is self-contained against external benchmarks. Thus no significant circularity.

Assumptions & free parameters 3 free parameters · 4 assumptions · 1 invented entities

The central claim rests on a hand-crafted Gaussian prior derived from ground-truth labels, an L1 pull toward that prior, and a low-rank Tucker regularizer whose ranks are unspecified. These constitute the main free choices and domain assumptions; the method introduces no new physical entities. The heaviest burden is the arbitrary variance rule and the reliance on the authors' own prior grid methods.

free parameters (3)
  • lambda (grid loss weight) = 0.5
    Chosen by hand in Eq. (9); controls the balance between label distribution loss and grid prior loss. No sensitivity analysis is provided.
  • Tucker ranks (R1, R2, R3) = not reported
    The low-rank constraint in Eq. (5) requires three rank values, but the paper never states how they are set; they control the denoising strength and are load-bearing for the method's claimed robustness.
  • Gaussian variance scaling 1 - |a_ij| = 1 - |a_ij|
    A design choice for the uncertainty of each grid cell in Section 3; the paper states that KL divergence and standard-normal priors were tried and did not work, but gives no quantitative comparison.
assumptions (4)
  • domain assumption Pairwise differences of ground-truth label degrees encode label correlation.
    The difference matrix D uses a_ij = d^j - d^i (Section 3), assuming arithmetic differences fully capture label relations.
  • ad hoc to paper Gaussian expansion with mean a_ij and variance 1 - |a_ij| models relation uncertainty.
    The prior grid B-hat is sampled from N(a_ij, 1 - |a_ij|) without derivation (Section 3).
  • domain assumption The learned grid should be close to the label-derived prior under L1 distance.
    Eq. (2) minimizes the L1 distance between B and B-hat, assuming the Gaussian-sampled prior is a good target representation.
  • domain assumption Tucker reconstruction with a low-rank assumption removes sampling noise.
    Eq. (5) applies Tucker decomposition and recovery to enforce low rank; the paper assumes this denoises the grid and improves prediction.
invented entities (1)
  • Latent Label Distribution Grid (LLDG)
    purpose: A c x c x c tensor representation that expands each label difference into a Gaussian distribution, used as the latent space for predicting label distributions.
    The LLDG is introduced solely as a model component and is validated only through the paper's own benchmarks; it provides no external falsifiable prediction.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Latent label distribution grid representation for modeling uncertainty." pith.science (2026). https://pith.science/paper/MEHZZW3U

@misc{pith2026250521180,
  author       = {Pith},
  title        = {Pith review of: Latent label distribution grid representation for modeling uncertainty},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/MEHZZW3U}},
  note         = {Machine review of arXiv:2505.21180}
}
read the original abstract

Although \textbf{L}abel \textbf{D}istribution \textbf{L}earning (LDL) has promising representation capabilities for characterizing the polysemy of an instance, the complexity and high cost of the label distribution annotation lead to inexact in the construction of the label space. The existence of a large number of inexact labels generates a label space with uncertainty, which misleads the LDL algorithm to yield incorrect decisions. To alleviate this problem, we model the uncertainty of label distributions by constructing a \textbf{L}atent \textbf{L}abel \textbf{D}istribution \textbf{G}rid (LLDG) to form a low-noise representation space. Specifically, we first construct a label correlation matrix based on the differences between labels, and then expand each value of the matrix into a vector that obeys a Gaussian distribution, thus building a LLDG to model the uncertainty of the label space. Finally, the LLDG is reconstructed by the LLDG-Mixer to generate an accurate label distribution. Note that we enforce a customized low-rank scheme on this grid, which assumes that the label relations may be noisy and it needs to perform noise-reduction with the help of a Tucker reconstruction technique. Furthermore, we attempt to evaluate the effectiveness of the LLDG by considering its generation as an upstream task to achieve the classification of the objects. Extensive experimental results show that our approach performs competitively on several benchmarks.

Figures

Figures reproduced from arXiv: 2505.21180 by the authors.

Figure 1
Figure 1. LLDG vs. label distribution matrix (LDM). This figure shows the representation patterns of LDM and LLDG, where LDM considers only the distribution of the values of the label distribution and LLDG considers the relational distribution of the values of the label distribution. between labels, which can help correct the problem of inaccurate labeling of a single label; second, make further processing of the representati… view at source ↗
Figure 2
Figure 2. Our architecture. In general, our model has two learning targets, on the one hand, to learn the label distribution, and on the other hand to learn the LLDG. Specifically, first, the input features are extracted by a local-global feature extractor to create a grid. LLDG establishes a labeled correlation space that is constrained by the Tucker reconstruction algorithm. Then, this grid is conducted in LLDG-Mixer to for… view at source ↗
Figure 3
Figure 3. This figure shows the energy distribution of the LLDG (4), and the blue boxes indicate the [PITH_FULL_IMAGE:figures/full_fig_p010_3.png] view at source ↗

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

25 extracted references · 25 canonical work pages

  1. [6]

    Label distribution learning via label correlation grid

    Qimeng Guo, Zhuoran Zheng, Xiuyi Jia, and Liancheng Xu. Label distribution learning via label correlation grid.arXiv preprint arXiv:2210.08184, 2022. 3

    work page Pith review arXiv 2022

  2. [25]

    Label Distribution Learning via Implicit Distribution Representation

    Zhuoran Zheng and Xiuyi Jia. Label distribution learning via implicit distribution representation. arXiv preprint arXiv:2209.13824, 2022. 1, 2, 3, 6 12 Checklist Please do not modify the questions and only use the provided macros for your answers. Note that the Checklist section does not count toward the page limit. In your paper, please delete this instr...

    work page Pith review arXiv 2022

  3. [1]

    Representation learning: A review and new perspectives.IEEE TPAMI, 2013

    Yoshua Bengio, Aaron Courville, and Pascal Vincent. Representation learning: A review and new perspectives.IEEE TPAMI, 2013. 3

    work page 2013

  4. [2]

    Jingying Chen, Chen Guo, Ruyi Xu, Kun Zhang, Zongkai Yang, and Honghai Liu. Toward children’s empathy ability analysis: Joint facial expression recognition and intensity estimation using label distribution learning.IEEE Transactions on Industrial Informatics, 2021. 3

    work page 2021

  5. [3]

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

    Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xiaohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias Minderer, Georg Heigold, Sylvain Gelly, et al. An image is worth 16x16 words: Transformers for image recognition at scale. InICLR, 2021. 4

    work page 2021

  6. [4]

    Age estimation using expectation of label distribution learning

    Bin-Bin Gao, Hong-Yu Zhou, Jianxin Wu, and Xin Geng. Age estimation using expectation of label distribution learning. InIJCAI, 2018. 2, 3

    work page 2018

  7. [5]

    Label distribution learning.IEEE TKDE, 2016

    Xin Geng. Label distribution learning.IEEE TKDE, 2016. 3, 6

    work page 2016

  8. [7]

    Label distribution learning with label correlations on local samples.IEEE TKDE, 2019

    Xiuyi Jia, Zechao Li, Xiang Zheng, Weiwei Li, and Sheng-Jun Huang. Label distribution learning with label correlations on local samples.IEEE TKDE, 2019. 2, 3

    work page 2019

Show all 25 references
  1. [8]

    Label distribution learning by maintaining label ranking relation.IEEE TKDE, 2021

    Xiuyi Jia, Xiaoxia Shen, Weiwei Li, Yunan Lu, and Jihua Zhu. Label distribution learning by maintaining label ranking relation.IEEE TKDE, 2021. 2, 3

    2021

  2. [9]

    Facial emotion distribu- tion learning by exploiting low-rank label correlations locally

    Xiuyi Jia, Xiang Zheng, Weiwei Li, Changqing Zhang, and Zechao Li. Facial emotion distribu- tion learning by exploiting low-rank label correlations locally. InCVPR, 2019. 3

    2019

  3. [10]

    Uncertainty- aware label distribution learning for facial expression recognition

    Nhat Le, Khanh Nguyen, Quang Tran, Erman Tjiputra, Bac Le, and Anh Nguyen. Uncertainty- aware label distribution learning for facial expression recognition. InWACV, 2023. 1

    2023

  4. [11]

    Unimodal-concentrated loss: Fully adaptive label distribution learning for ordinal regression

    Qiang Li, Jingjing Wang, Zhaoliang Yao, Yachun Li, Pengju Yang, Jingwei Yan, Chunmao Wang, and Shiliang Pu. Unimodal-concentrated loss: Fully adaptive label distribution learning for ordinal regression. InCVPR, 2022. 1, 3

    2022

  5. [12]

    Label distribution learning with noisy labels via three-way decisions.International Journal of Approximate Reasoning, 2022

    Weiwei Li, Yuqing Lu, Lei Chen, and Xiuyi Jia. Label distribution learning with noisy labels via three-way decisions.International Journal of Approximate Reasoning, 2022. 1, 3

    2022

  6. [13]

    Feature selection for label dis- tribution learning via feature similarity and label correlation.Information Sciences, 2022

    Wenbin Qian, Yinsong Xiong, Jun Yang, and Wenhao Shu. Feature selection for label dis- tribution learning via feature similarity and label correlation.Information Sciences, 2022. 3

    2022

  7. [14]

    Label distribution learning with label-specific features

    Tingting Ren, Xiuyi Jia, Weiwei Li, Lei Chen, and Zechao Li. Label distribution learning with label-specific features. InIJCAI, 2019. 2, 3

    2019

  8. [15]

    Label distribution learning with label correlations via low-rank approximation

    Tingting Ren, Xiuyi Jia, Weiwei Li, and Shu Zhao. Label distribution learning with label correlations via low-rank approximation. InIJCAI, 2019. 2, 3

    2019

  9. [16]

    Towards speaker age estimation with label distribution learning

    Shijing Si, Jianzong Wang, Junqing Peng, and Jing Xiao. Towards speaker age estimation with label distribution learning. InICASSP, 2022. 3

    2022

  10. [17]

    Co-ldl: A co-training-based label distribution learning method for tackling label noise.IEEE Transactions on Multimedia, 2021

    Zeren Sun, Huafeng Liu, Qiong Wang, Tianfei Zhou, Qi Wu, and Zhenmin Tang. Co-ldl: A co-training-based label distribution learning method for tackling label noise.IEEE Transactions on Multimedia, 2021. 1

    2021

  11. [18]

    A label distribution manifold learning algorithm.PR, 2022

    Chao Tan, Sheng Chen, Xin Geng, and Genlin Ji. A label distribution manifold learning algorithm.PR, 2022. 3

    2022

  12. [19]

    A label distribution manifold learning algorithm.PR, 2023

    Chao Tan, Sheng Chen, Xin Geng, and Genlin Ji. A label distribution manifold learning algorithm.PR, 2023. 1

    2023

  13. [20]

    Incomplete label distribution learning by exploiting global sample correlation

    Qifa Teng and Xiuyi Jia. Incomplete label distribution learning by exploiting global sample correlation. InMultimedia Understanding with Less Labeling on Multimedia Understanding with Less Labeling. 2021. 3 11

    2021

  14. [21]

    Label distribution learning by exploiting label distribution manifold

    Jing Wang and Xin Geng. Label distribution learning by exploiting label distribution manifold

  15. [22]

    Multi-class ASD classification via label distribution learning with class-shared and class-specific decomposition.Medical Image Analysis, 2022

    Jun Wang, Fengyexin Zhang, Xiuyi Jia, Xin Wang, Han Zhang, Shihui Ying, Qian Wang, Jun Shi, and Dinggang Shen. Multi-class ASD classification via label distribution learning with class-shared and class-specific decomposition.Medical Image Analysis, 2022. 1

    2022

  16. [23]

    Label enhancement for label distribution learning

    Ning Xu, Yun-Peng Liu, and Xin Geng. Label enhancement for label distribution learning. IEEE TKDE, 2019. 1

    2019

  17. [24]

    Robust lightweight facial expression recognition network with label distribution training

    Zengqun Zhao, Qingshan Liu, and Feng Zhou. Robust lightweight facial expression recognition network with label distribution training. InAAAI, 2021. 3

    2021

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.