Pith. sign in

REVIEW 4 major objections 3 minor 41 references

Reweighting classes by real-time learning progress, not static counts, boosts long-tailed accuracy.

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 ↗

T0 review · deepseek-v4-flash

2026-08-01 01:20 UTC pith:6EEVNOAV

load-bearing objection A plausible plug-and-play dynamic loss for long-tailed classification, with honest ablations and stability checks, but the empirical claims are weakened by test-set-tuned hyperparameters, best-of-runs reporting, and an unexamined normalizer in the loss. the 4 major comments →

arxiv 2607.25830 v1 pith:6EEVNOAV submitted 2026-07-28 cs.CV

Beyond Static Costs: Learning-Dynamics Aware Loss Functions for Long-Tailed Classification

Varad Shinde , Nikhil Kumar Shrey , Magesh Rajasekaran , Md Saiful Islam Sajol , Harshil Bhargava , Subhajit Sidanta , Supratik Mukhopadhyay , Yimin Zhu This is my paper
classification cs.CV
keywords long-tailed classificationreweighting losslearning dynamicsclass imbalanceShannon entropysemantic scaleinter-epoch regularizationsoft prediction counts
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 paper is trying to establish that loss reweighting for long-tailed classification should respond to how well each class is being learned during training, not to fixed sample frequencies. It proposes LDAL, a drop-in auxiliary loss that up-weights classes whose soft prediction counts are high relative to their current difficulty, where difficulty is measured by prediction entropy and feature-strength (semantic scale). On CIFAR-10-LT, CIFAR-100-LT, ImageNet-LT, and iNaturalist-2018, LDAL reports top-1 accuracy above the prior best pure reweighting loss, e.g., 45.04 vs. 43.85 on CIFAR-100-LT (ρ=200) and 50.10 vs. 49.53 on ImageNet-LT. If correct, this shows that objective-level dynamic weighting can capture much of the benefit of module-heavy long-tailed methods while remaining a simple plug-in.

Core claim

On the paper's own terms, the central discovery is that a loss whose per-class penalty is a function of dynamic learning state—soft prediction counts, prediction entropy, semantic scale, and an inter-epoch prediction-shift regularizer—consistently outperforms static inverse-frequency and margin-based reweighting losses across four long-tailed image benchmarks. The penalty for class i is ((γ_i * soft-count_i + r_i)^2) divided by a stop-gradient normalizer that measures mean squared deviation between the batch logits and the one-hot vector of class i. The dynamic weight γ_i is built so that classes that are rare but already well learned (low entropy, high semantic scale) are not heavily penali

What carries the argument

The load-bearing object is the LDAL auxiliary loss term in Eq. (1): for each class i, L_i = (γ_i * \tilde{N}_i + r_i)^2 / ( (1/(BC)) Σ_{b,k}(sg(z_k^{(b)}) - (e_i)_k)^2 + ε ), where \tilde{N}_i is a differentiable soft prediction count, γ_i = min( S_i / ( (1 + max_j S_j) * H_i ), τ ) combines semantic scale S_i (squared mean feature logit norm) with mean prediction entropy H_i, r_i is an inter-epoch regularizer based on hard argmax count changes for target minority classes, and the stop-gradient denominator is asserted to keep penalty scale consistent. This mechanism carries the argument by converting static class-frequency priors into epoch-level learning-progress signals while retaining gra

Load-bearing premise

The stop-gradient denominator in Eq. (1) is assumed to be a harmless scale-consistent normalizer, but the paper never derives or ablates it; if it couples class i's penalty to the logits of all samples in ways that skew relative weights, the reported gains could be an artifact of the normalizer rather than of dynamic weighting.

What would settle it

Train on CIFAR-100-LT with the LDAL denominator replaced by (i) a constant 1.0, (ii) a per-class stop-gradient distance to a fixed centroid, and (iii) a detached version of the original denominator. If top-1 accuracy stays within run-to-run noise across all three, the denominator is inert; if accuracy drops sharply when the denominator is constant or when the stop-gradient is removed, the denominator is the carrier of the effect.

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

If this is right

  • LDAL can be appended to any classification objective with negligible overhead, so dynamic reweighting becomes a drop-in change rather than a new architecture or training schedule.
  • If the reported accuracy holds across five seeds, LDAL sets a new reference point among pure class-sensitive losses, closing part of the gap to decoupled and ensemble methods without multi-stage training.
  • The inter-epoch regularizer's benefit appears concentrated after learning-rate decay, suggesting dynamic reweighting methods should pay attention to late-stage convergence, not just early rebalancing.
  • The gradient-flow analysis shows the auxiliary term's relative contribution grows as training matures, implying the method's mechanism is strongest during fine-grained feature learning.
  • The paper's proposed extension to object detection and transformer backbones would test whether the same learning-dynamics signals transfer outside the evaluated CNN classification setting.

Where Pith is reading between the lines

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

  • A cheap falsification test would replace the stop-gradient denominator with a per-class constant and re-run the benchmarks; if accuracy is unchanged, the denominator is inert and the gains come from the numerator terms alone.
  • The reported Spearman ρ=-0.94 between training-time entropy and final test accuracy suggests entropy could be repurposed as an early-stopping or sample-scheduling signal beyond its role in the loss.
  • The α-sweep shows instability on CIFAR-10 at α>5, hinting that the regularizer's safe range may shrink as the number of classes grows; a class-count-aware α schedule would be a natural extension the paper does not explore.
  • Since LDAL is purely objective-level, a head-to-head against sampling-based hybrids under identical backbones would clarify whether dynamic weighting subsumes sampling or they are complementary.

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 / 3 minor

Summary. The paper proposes LDAL, an auxiliary loss for long-tailed classification. In addition to cross-entropy, LDAL adds a per-class penalty (γ_i \tilde N_i + r_i)^2 divided by a stop-gradient denominator (Eq. 1). γ_i combines a logit-norm 'semantic scale' S_i and prediction entropy H_i (Eqs. 2–4), while r_i is an inter-epoch regularizer based on hard prediction-count changes for minority classes (Eq. 5). Experiments on CIFAR-10/100-LT, ImageNet-LT, and iNaturalist-2018 report accuracy gains over AREA and other CSL baselines, together with ablations and gradient-flow diagnostics.

Significance. If the reported gains hold under a properly controlled evaluation, LDAL would be a valuable plug-and-play objective for long-tailed classification: it operates purely at the loss level, ships with code, and includes component ablations and stability analyses. However, the central claim currently rests on comparisons that are difficult to interpret, and the most novel term in the loss—the stop-gradient denominator—is neither derived nor ablated. The paper's explanatory narrative could therefore be misleading if that denominator, rather than the semantic-scale/entropy/regularizer signals, drives the improvements.

major comments (4)
  1. [Section II-B, Eq. (1)] The denominator D_i = (1/(BC)) Σ_{b,k}(sg(z_k^b)-(e_i)_k)^2 is introduced only with the assertion that it 'keeps the penalty scale consistent.' It is not a constant normalizer: D_i is a per-class, batch-dependent function of all current logits. Expanding for class i gives D_i = (1/B) Σ_b (Σ_{k≠i} z_{k,b}^2 + (z_{i,b}-1)^2), so classes with large logit magnitudes receive smaller penalties, and the effect changes over training. This term is never derived, ablated, or given a sensitivity analysis. If the reported gains are driven by this denominator rather than by S_i, H_i, or r_i, the interpretation in the paper is misleading. Please analyze the denominator's behavior, ablate it by replacing D_i with a constant or a per-class logit-norm normalizer, and report the resulting accuracy.
  2. [Section V-C / Table VIII] The hyperparameter α is selected by sweeping on the same CIFAR benchmarks used for the main results: α=1.0 is chosen from Table VIII and then used in Table I. No held-out validation split or cross-validation procedure is described. This amounts to test-set selection for a free parameter and can inflate the reported gains. Please either tune α on a separate validation split and report those numbers, or show that the Table I conclusions are unchanged when α is fixed to a value chosen without access to the test set.
  3. [Tables I and III (footnotes)] LDAL results are reported as the 'best top-1 accuracy across all runs,' whereas the baselines are point estimates taken from prior papers. Table V shows that mean LDAL accuracy is lower than the headline numbers: CIFAR-100 ρ=100 mean 48.88 vs. best 49.79; ImageNet-LT mean 49.67 vs. best 50.10; iNaturalist mean 66.53 vs. best 67.10. Comparing best-of-runs against literature point estimates biases the comparison in LDAL's favor. Please report mean±std for LDAL and, where possible, for baselines, and state whether the differences are statistically significant.
  4. [Section IV-A / IV-B] The training protocol for baselines is not fully specified: the paper states that implementation details were 'adopted from AREA [4],' but does not list epochs, learning-rate schedules, augmentations, batch sizes, or the exact values of τ, α, and the target minority set U used per dataset. This makes it hard to verify that the comparison is fair and to reproduce the results. Please provide a protocol table and release configuration files for all methods.
minor comments (3)
  1. [Various] Typos: 'eliminting' (Section IV-B), 'throughtout' (Section V-A), 'Remooving' (Section V-B). Section IV-B refers to 'Section 3.3' instead of the numbered section for γ_i.
  2. [Eq. (4) / Section II-D] The upper bound τ=5.0 and the stability constant ε=1.0 are fixed without sensitivity analysis or justification. The target minority class set U is also not defined in a reproducible way.
  3. [Table V] The 95% confidence intervals appear to use z=1.96 rather than a t-distribution for n=5. The intervals would be wider with the correct t-quantile (2.776).

Circularity Check

0 steps flagged

No significant circularity: LDAL is an empirical loss-design proposal whose benchmarks are external and whose free parameters are selected by sensitivity analysis, not derived from the target result.

full rationale

The central claim is that the proposed LDAL objective outperforms existing reweighting losses on standard long-tailed benchmarks. This is an empirical claim supported by comparisons to published methods (AREA, LDAM, IB, etc.) on CIFAR, ImageNet-LT, and iNaturalist-2018; no quantity in the loss is fit to those test accuracies in a way that would force the reported numbers. The loss components (semantic scale, entropy-based difficulty, inter-epoch regularizer, and the stop-gradient denominator) are presented as design choices with heuristic rationales, not as predictions derived from first principles or from a self-referential definition. The denominator in Eq. (1) is indeed never ablated and could plausibly influence results, but that is a missing-analysis / robustness concern, not a circular step: it does not reduce the reported outcome to an input by construction. Hyperparameters α and τ are tuned via sensitivity analyses on validation performance, which is ordinary hyperparameter selection rather than fitted-input-called-prediction. The paper cites prior work for concepts like 'semantic values' and category information amount, but those citations are external and are used as motivation, not as the sole justification of the central result. The only self-citations ([2], [5], [32], [33]) appear in contextual remarks about satellite imaging and service composition and are not load-bearing for the proposed loss. No uniqueness theorem is imported from the authors' prior work, and no known result is merely renamed as a new framework. The derivation chain is therefore self-contained with respect to the empirical comparisons; the main risks are methodological (unablated denominator, best-run reporting) rather than circular.

Axiom & Free-Parameter Ledger

4 free parameters · 3 axioms · 0 invented entities

The central claim rests on several assumptions: H_i as intrinsic difficulty, S_i as feature strength, and the denominator as a valid normalizer. The free parameters τ and α are fixed; α was tuned on test data. The target minority set U is unspecified.

free parameters (4)
  • τ (gamma upper bound) = 5.0
    Set by hand as a stability cap on the dynamic importance weight γ_i; no sensitivity analysis reported.
  • α (inter-epoch regularizer strength) = 1.0
    Selected as the best value from a sweep over α∈{0.5,1.0,2.0,5.0,10.0,20.0} on the same CIFAR benchmarks used for headline results (Table VIII); this is test-set tuning.
  • ε (stability constant) = 1.0
    Chosen by hand to prevent division by zero; no sensitivity analysis.
  • target minority class set U = unspecified
    The set of classes receiving the inter-epoch regularizer is never defined algorithmically; presumably chosen per dataset by the authors.
axioms (3)
  • domain assumption Per-class prediction entropy H_i is a stable, meaningful measure of intrinsic class difficulty (easy classes are confidently predicted).
    The paper offers a correlation analysis (Fig. 3) on the same benchmark, but entropy is measured online from at most a few samples per tail class per batch, so its stability is not established.
  • domain assumption Semantic scale S_i, the squared mean logit norm, measures feature representation strength and is larger for head classes.
    Borrowed from prior work [19]; the authors do not verify this in their own settings beyond the ablation.
  • ad hoc to paper The stop-gradient denominator D_i in Eq. 1 acts as a harmless scale-consistent normalizer.
    Introduced without derivation or ablation; the central reweighting behavior depends on it.

pith-pipeline@v1.3.0-alltime-deepseek · 11940 in / 15926 out tokens · 131190 ms · 2026-08-01T01:20:55.674532+00:00 · methodology

0 comments
read the original abstract

Deep learning models in computer vision face significant challenges when trained on long-tailed datasets, where a few majority classes dominate while many minority classes are severely underrepresented. Such imbalances frequently arise in real-world scenarios such as rare species recognition, manufacturing fault detection, and medical image understanding, leading to biased models that underperform on tail classes. Existing reweighting methods typically rely on static class frequencies to penalize the model, ignoring the dynamic nature of how effectively a network actually learns a class over time. We address this by introducing a novel Learning-Dynamics Aware Loss (LDAL) function that shifts the focus from static sample counts to dynamic learning progress. LDAL framework adjusts class weights continuously by leveraging: (i) the strength of learned feature representations (semantic scale), (ii) the intrinsic learning difficulty of each class, measured via the Shannon entropy of its predictions, and (iii) an inter-epoch regularizer term that tracks prediction shifts between consecutive epochs to stabilize training and avoid local minima. LDAL is purely a objective function which incurs negligible computational overhead while adapting to the feature learning of the model. Experimental results on multiple benchmark datasets demonstrate that our approach significantly surpasses state-of-the-art reweighting loss functions, providing an optimal trade-off between accuracy and generalizability. The source code is available at https://github.com/sdm2026/ldal

Figures

Figures reproduced from arXiv: 2607.25830 by Harshil Bhargava, Magesh Rajasekaran, Md Saiful Islam Sajol, Nikhil Kumar Shrey, Subhajit Sidanta, Supratik Mukhopadhyay, Varad Shinde, Yimin Zhu.

Figure 1
Figure 1. Figure 1: Overview of the LDAL penalty computation. [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Validation accuracy curves comparing LDAL with [PITH_FULL_IMAGE:figures/full_fig_p007_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: shows a strong negative correlation (Spearman ρ = −0.94, p < 10−4 ): classes with higher prediction entropy during training consistently achieve lower final accuracy. Notably, visually simple classes such as Airplane and Car exhibit low entropy and high accuracy, while semantically complex classes like Dog and Horse exhibit high entropy and low accuracy—independent of their position in the long￾tailed dist… 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

41 extracted references · 1 linked inside Pith

  1. [1]

    https://github.com/sdm2026/ldal

    Ldal github repository. https://github.com/sdm2026/ldal

  2. [2]

    Deepsat: a learning framework for satellite imagery

    Saikat Basu, Sangram Ganguly, Supratik Mukhopadhyay, Robert DiB- iano, Manohar Karki, and Ramakrishna Nemani. Deepsat: a learning framework for satellite imagery. InProceedings of the 23rd SIGSPATIAL international conference on advances in geographic information systems, pages 1–10, 2015

    2015

  3. [3]

    Learning imbalanced datasets with label-distribution-aware margin loss, 2019

    Kaidi Cao, Colin Wei, Adrien Gaidon, Nikos Arechiga, and Tengyu Ma. Learning imbalanced datasets with label-distribution-aware margin loss, 2019

    2019

  4. [4]

    Area: Adaptive reweighting via effective area for long-tailed classification

    Xiaohua Chen, Yucan Zhou, Dayan Wu, Chule Yang, Bo Li, Qinghua Hu, and Weiping Wang. Area: Adaptive reweighting via effective area for long-tailed classification. In2023 IEEE/CVF International Conference on Computer Vision (ICCV), pages 19220–19230, 2023. 8

    2023

  5. [5]

    Progressively growing generative adversarial networks for high resolution semantic segmentation of satellite images

    Edward Collier, Kate Duffy, Sangram Ganguly, Geri Madanguit, Subodh Kalia, Gayaka Shreekant, Ramakrishna Nemani, Andrew Michaelis, Shaung Li, Auroop Ganguly, et al. Progressively growing generative adversarial networks for high resolution semantic segmentation of satellite images. In2018 IEEE International Conference on Data Mining Workshops (ICDMW), page...

    2018

  6. [6]

    Class-balanced loss based on effective number of samples, 2019

    Yin Cui, Menglin Jia, Tsung-Yi Lin, Yang Song, and Serge Belongie. Class-balanced loss based on effective number of samples, 2019

    2019

  7. [7]

    Class rectification hard mining for imbalanced deep learning

    Qi Dong, Shaogang Gong, and Xiatian Zhu. Class rectification hard mining for imbalanced deep learning. InProceedings of the IEEE international conference on computer vision, pages 1851–1860, 2017

    2017

  8. [8]

    Borderline-smote: a new over-sampling method in imbalanced data sets learning

    Hui Han, Wen-Yuan Wang, and Bing-Huan Mao. Borderline-smote: a new over-sampling method in imbalanced data sets learning. InInternational conference on intelligent computing, pages 878–887. Springer, 2005

    2005

  9. [9]

    Learning deep representation for imbalanced classification

    Chen Huang, Yining Li, Chen Change Loy, and Xiaoou Tang. Learning deep representation for imbalanced classification. InProceedings of the IEEE conference on computer vision and pattern recognition, pages 5375–5384, 2016

    2016

  10. [10]

    Decoupling representation and classifier for long-tailed recognition, 2020

    Bingyi Kang, Saining Xie, Marcus Rohrbach, Zhicheng Yan, Albert Gordo, Jiashi Feng, and Yannis Kalantidis. Decoupling representation and classifier for long-tailed recognition, 2020

    2020

  11. [11]

    M2m: Imbalanced classification via major-to-minor translation

    Jaehyung Kim, Jongheon Jeong, and Jinwoo Shin. M2m: Imbalanced classification via major-to-minor translation. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 13896–13905, 2020

    2020

  12. [12]

    Re-weighted softmax cross-entropy to control forgetting in federated learning, 2023

    Gwen Legate, Lucas Caccia, and Eugene Belilovsky. Re-weighted softmax cross-entropy to control forgetting in federated learning, 2023

    2023

  13. [13]

    Focal-sam: Focal sharpness-aware minimization for long-tailed classification, 2025

    Sicong Li, Qianqian Xu, Zhiyong Yang, Zitai Wang, Linchao Zhang, Xiaochun Cao, and Qingming Huang. Focal-sam: Focal sharpness-aware minimization for long-tailed classification, 2025

    2025

  14. [14]

    Focal loss for dense object detection

    Tsung-Yi Lin, Priya Goyal, Ross Girshick, Kaiming He, and Piotr Dollár. Focal loss for dense object detection. InProceedings of the IEEE international conference on computer vision, pages 2980–2988, 2017

    2017

  15. [15]

    Focal loss for dense object detection, 2018

    Tsung-Yi Lin, Priya Goyal, Ross Girshick, Kaiming He, and Piotr Dollár. Focal loss for dense object detection, 2018

    2018

  16. [16]

    Large-scale long-tailed recognition in an open world

    Ziwei Liu, Zhongqi Miao, Xiaohang Zhan, Jiayun Wang, Boqing Gong, and Stella X Yu. Large-scale long-tailed recognition in an open world. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 2537–2546, 2019

    2019

  17. [17]

    Ziwei Liu, Zhongqi Miao, Xiaohang Zhan, Jiayun Wang, Boqing Gong, and Stella X. Yu. Large-scale long-tailed recognition in an open world, 2019

    2019

  18. [18]

    Pursuing better decision boundaries for long-tailed object detection via category information amount, 2025

    Yanbiao Ma, Wei Dai, and Jiayi Chen. Pursuing better decision boundaries for long-tailed object detection via category information amount, 2025

    2025

  19. [19]

    Delving into semantic scale imbalance, 2023

    Yanbiao Ma, Licheng Jiao, Fang Liu, Yuxin Li, Shuyuan Yang, and Xu Liu. Delving into semantic scale imbalance, 2023

    2023

  20. [20]

    Factors in finetuning deep model for object detection with long-tail distribution

    Wanli Ouyang, Xiaogang Wang, Cong Zhang, and Xiaokang Yang. Factors in finetuning deep model for object detection with long-tail distribution. InProceedings of the IEEE conference on computer vision and pattern recognition, pages 864–873, 2016

    2016

  21. [21]

    The majority can help the minority: Context-rich minority oversampling for long-tailed classification

    Seulki Park, Youngkyu Hong, Byeongho Heo, Sangdoo Yun, and Jin Young Choi. The majority can help the minority: Context-rich minority oversampling for long-tailed classification. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 6887–6896, 2022

    2022

  22. [22]

    Influence- balanced loss for imbalanced visual classification, 2021

    Seulki Park, Jongin Lim, Younghan Jeon, and Jin Young Choi. Influence- balanced loss for imbalanced visual classification, 2021

    2021

  23. [23]

    Balanced meta-softmax for long-tailed visual recognition, 2020

    Jiawei Ren, Cunjun Yu, Shunan Sheng, Xiao Ma, Haiyu Zhao, Shuai Yi, and Hongsheng Li. Balanced meta-softmax for long-tailed visual recognition, 2020

    2020

  24. [24]

    Managing bias in ai

    Drew Roselli, Jeanna Matthews, and Nisha Talagala. Managing bias in ai. InCompanion proceedings of the 2019 world wide web conference, pages 539–544, 2019

    2019

  25. [25]

    Lift+: Lightweight fine-tuning for long-tail learning, 2025

    Jiang-Xin Shi, Tong Wei, and Yu-Feng Li. Lift+: Lightweight fine-tuning for long-tail learning, 2025

    2025

  26. [26]

    Meta-weight-net: Learning an explicit mapping for sample weighting

    Jun Shu, Qi Xie, Lixuan Yi, Qian Zhao, Sanping Zhou, Zongben Xu, and Deyu Meng. Meta-weight-net: Learning an explicit mapping for sample weighting. In H. Wallach, H. Larochelle, A. Beygelzimer, F. d'Alché- Buc, E. Fox, and R. Garnett, editors,Advances in Neural Information Processing Systems, volume 32. Curran Associates, Inc., 2019

    2019

  27. [27]

    Solar: Sinkhorn label refinery for imbalanced partial-label learning, 2022

    Haobo Wang, Mingxuan Xia, Yixuan Li, Yuren Mao, Lei Feng, Gang Chen, and Junbo Zhao. Solar: Sinkhorn label refinery for imbalanced partial-label learning, 2022

    2022

  28. [28]

    Rsg: A simple but effective module for learning imbalanced datasets, 2021

    Jianfeng Wang, Thomas Lukasiewicz, Xiaolin Hu, Jianfei Cai, and Zhenghua Xu. Rsg: A simple but effective module for learning imbalanced datasets, 2021

    2021

  29. [30]

    Xudong Wang, Long Lian, Zhongqi Miao, Ziwei Liu, and Stella X. Yu. Long-tailed recognition by routing diverse distribution-aware experts. CoRR, abs/2010.01809, 2020

    Pith/arXiv arXiv 2010

  30. [31]

    A survey on long- tailed visual recognition.International Journal of Computer Vision, 130(7):1837–1872, 2022

    Lu Yang, He Jiang, Qing Song, and Jun Guo. A survey on long- tailed visual recognition.International Journal of Computer Vision, 130(7):1837–1872, 2022

    2022

  31. [32]

    Automated situation- aware service composition in service-oriented computing.International Journal of Web Services Research (IJWSR), 4(4):59–82, 2007

    Stephen S Yau, Hasan Davulcu, Supratik Mukhopadhyay, Dazhi Huang, Haishan Gong, Prabhdeep Singh, and Fatih Gelgi. Automated situation- aware service composition in service-oriented computing.International Journal of Web Services Research (IJWSR), 4(4):59–82, 2007

    2007

  32. [33]

    Adaptable situation-aware secure service-based (as/sup 3/) systems

    Stephen S Yau, Hasan Davulcu, Supratik Mukhopadhyay, Dazhi Huang, and Yisheng Yao. Adaptable situation-aware secure service-based (as/sup 3/) systems. InEighth IEEE International Symposium on Object-Oriented Real-Time Distributed Computing (ISORC’05), pages 308–315. IEEE, 2005

    2005

  33. [34]

    Self- supervised aggregation of diverse experts for test-agnostic long-tailed recognition.Advances in Neural Information Processing Systems, 35:34077–34090, 2022

    Yifan Zhang, Bryan Hooi, Lanqing Hong, and Jiashi Feng. Self- supervised aggregation of diverse experts for test-agnostic long-tailed recognition.Advances in Neural Information Processing Systems, 35:34077–34090, 2022

    2022

  34. [35]

    Self- supervised aggregation of diverse experts for test-agnostic long-tailed recognition, 2022

    Yifan Zhang, Bryan Hooi, Lanqing Hong, and Jiashi Feng. Self- supervised aggregation of diverse experts for test-agnostic long-tailed recognition, 2022

    2022

  35. [36]

    Deep long-tailed learning: A survey, 2023

    Yifan Zhang, Bingyi Kang, Bryan Hooi, Shuicheng Yan, and Jiashi Feng. Deep long-tailed learning: A survey, 2023

    2023

  36. [37]

    Zhilu Zhang and Mert R. Sabuncu. Generalized cross entropy loss for training deep neural networks with noisy labels, 2018

    2018

  37. [38]

    Learning fast sample re-weighting without reward data, 2021

    Zizhao Zhang and Tomas Pfister. Learning fast sample re-weighting without reward data, 2021

    2021

  38. [39]

    Improving calibration for long-tailed recognition

    Zhisheng Zhong, Jiequan Cui, Shu Liu, and Jiaya Jia. Improving calibration for long-tailed recognition. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 16489– 16498, 2021

    2021

  39. [40]

    Improving calibration for long-tailed recognition, 2021

    Zhisheng Zhong, Jiequan Cui, Shu Liu, and Jiaya Jia. Improving calibration for long-tailed recognition, 2021

    2021

  40. [41]

    Bbn: Bilateral-branch network with cumulative learning for long-tailed visual recognition

    Boyan Zhou, Quan Cui, Xiu-Shen Wei, and Zhao-Min Chen. Bbn: Bilateral-branch network with cumulative learning for long-tailed visual recognition. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 9719–9728, 2020

    2020

  41. [42]

    Inflated episodic memory with region self-attention for long-tailed visual recognition

    Linchao Zhu and Yi Yang. Inflated episodic memory with region self-attention for long-tailed visual recognition. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 4344–4353, 2020

    2020