- Rating
- 3
- Confidence
- 3(+1)
- Informativeness
- 3
- Presentation
- 2
- Soundness
- 2
Read review
Paper summary
This paper studies transductive few-shot clustering with labels covering only a subset of categories. DD-FSC combines a noise- conditional score model with Dirichlet-process mixture inference. Score-guided feature trajectories refine semantic anchors, expand the labeled groups with confident unlabeled samples, and provide targets for representation learning. The inferred cluster structure feeds back into the trajectories and a prototype-level objective. The framework supports both known and adaptively inferred cluster counts.
Critical strengths
-
The paper is well written and well structured.
-
A useful framework for partial category coverage. The method connects sparse semantic supervision, density-based representation refinement, and adaptive cluster discovery. This addresses a meaningful setting where some categories have no labeled examples and the cluster count may be unknown. The feedback between representation learning and cluster inference is a useful aspect of the design.
-
Broad empirical support for the combined framework. The evaluation covers multiple image benchmarks, both known- and unknown-K settings, and generalized category discovery. The ablations report lower performance when score guidance or DP–diffusion coupling is removed. These results support the practical value of the components within the proposed framework.
Critical weaknesses
-
The contribution over prior density-based methods needs clearer discussion. DIPO [1] and TAFSSL [2] refine few-shot prototypes using unlabeled neighborhoods. DIPO does so without additional model training. CMS [3] combines mean-shift refinement with contrastive learning for category discovery. HG-VAE [4] uses high- density hub samples to improve feature generation. CMS is cited and evaluated, but the added benefit of learned score guidance remains unclear. The kNN expansion ablation does not isolate this benefit over DIPO-style refinement. A matched comparison would help clarify the performance gains and additional training cost.
-
The effect of unlabeled class imbalance needs further evaluation. Prior work shows that transductive few-shot gains can depend strongly on balanced query sets [5]. The ImageNet-50/100/200 experiments use balanced subsets. Appendix G.2 varies support counts rather than unlabeled class proportions. The Herbarium19 results are useful, but use the GCD supervision protocol. A controlled imbalance study under the main few-shot setting would help assess whether the gains persist, especially when K is unknown.
[1] Li et al. A Density-driven Iterative Prototype Optimization for Transductive Few-shot Learning (https://www.ijcai.org/proceedings/2024/0488.pdf). IJCAI, 2024.
[2] Lichtenstein et al. TAFSSL: Task-Adaptive Feature Sub-Space Learning for Few-Shot Classification (https://arxiv.org/abs/2003.06670). ECCV, 2020.
[3] Choi et al. Contrastive Mean-Shift Learning for Generalized Category Discovery (https://openaccess.thecvf.com/content/CVPR2024/papers/Choi_Contrastive_Mean-Shift_Learning_for_Generalized_Category_Discovery_CVPR_2024_paper.pdf). CVPR, 2024.
[4] Chen et al. A Novel Perspective on Leveraging Hubness in VAE for Eliminating Representative Shift Vectors in Few-Shot Learning (https://doi.org/10.1109/ICME59968.2025.11209098). ICME, 2025.
[5] Veilleux et al. Realistic Evaluation of Transductive Few-Shot Learning (https://proceedings.neurips.cc/paper_files/paper/2021/file/4d7a968bb636e25818ff2a3941db08c1-Paper.pdf). NeurIPS, 2021.
Questions for authors
Please see Critical Weaknesses.
5 comments
W2-1. Unlabeled imbalance needs controlled evaluation under the main few-shot protocol, especially with unknown ; ImageNet subsets are balanced, and Appendix G.2 varies supports rather than unlabeled proportions.
R2-1. We added an unlabeled class imbalance study under the main five-shot protocol, covering both known- and unknown- settings. To isolate imbalance, paired runs keep image counts, supports, and training and prediction settings fixed, varying only unlabeled class proportions. Following the Dirichlet sampling principle in the work you recommended, Veilleux et al., we draw proportions from . Balanced pools contain equal numbers per class; Dirichlet imbalanced pools follow sampled proportions while preserving the total count and retaining every class. DD-FSC is retrained in both settings; main-paper results serve only as a reference. Metrics are three-run means; estimated counts are mean standard deviation.
| (a) Known- | ||||
|---|---|---|---|---|
| Dataset | Setting | NMI (%) | ACC (%) | ARI (%) |
| ImageNet-50 | Main-paper results | 95.5 | 93.4 | 87.3 |
| Balanced | 93.8 | 91.2 | 84.1 | |
| Dirichlet imbalanced | 91.4 | 89.4 | 84.4 | |
| ImageNet-100 | Main-paper results | 92.4 | 85.2 | 77.4 |
| Balanced | 90.7 | 83.5 | 76.1 | |
| Dirichlet imbalanced | 88.3 | 80.4 | 73.2 | |
| ImageNet-200 | Main-paper results | 88.7 | 83.7 | 71.4 |
| Balanced | 87.0 | 81.3 | 70.9 | |
| Dirichlet imbalanced | 85.5 | 78.0 | 69.2 | |
| ImageNet-1K | Main-paper results | 86.4 | 72.4 | 51.2 |
| Balanced | 85.9 | 70.4 | 48.3 | |
| Dirichlet imbalanced | 83.5 | 68.1 | 46.1 |
| (b) Unknown- | |||||
|---|---|---|---|---|---|
| Dataset | Setting | NMI (%) | ACC (%) | ARI (%) | |
| ImageNet-50 | Main-paper results | 82.9 | 76.0 | 66.8 | |
| Balanced | 81.2 | 73.9 | 64.7 | ||
| Dirichlet imbalanced | 78.9 | 71.7 | 62.3 | ||
| ImageNet-100 | Main-paper results | 69.7 | 56.2 | 44.7 | |
| Balanced | 67.7 | 54.3 | 42.5 | ||
| Dirichlet imbalanced | 65.4 | 51.6 | 39.4 | ||
| ImageNet-200 | Main-paper results | 63.6 | 46.7 | 35.3 | |
| Balanced | 62.1 | 45.8 | 33.2 | ||
| Dirichlet imbalanced | 59.7 | 41.5 | 30.1 | ||
| ImageNet-1K | Main-paper results | 60.5 | 36.6 | 28.1 | |
| Balanced | 57.4 | 34.7 | 26.8 | ||
| Dirichlet imbalanced | 56.7 | 31.3 | 23.3 |
Compared with Balanced, Dirichlet imbalanced reduces ACC by 1.8–3.3 percentage points with known and 2.2–4.3 points with unknown . Mean estimated counts remain within 8% of the true count on all four imbalanced datasets. These results show that DD-FSC remains effective beyond balanced unlabeled sets, with limited accuracy declines and near-correct category counts under the tested imbalance.
W2-2. The Herbarium19 results are useful, but use GCD supervision rather than the main few-shot protocol.
R2-2. We also added unknown- few-shot experiments on Herbarium19 in Appendix D.4, using its naturally unequal class sizes. This tests performance with both abundant and scarce categories. The table combines clustering metrics and estimated counts from one seed; the true count is 683.
| Method | NMI (%) | ACC (%) | ARI (%) | |
|---|---|---|---|---|
| Soft -Means (ICLR, 2018) | 45.4 | 27.9 | 27.9 | 605 |
| cluster-FSL (CVPR, 2022) | 50.0 | 32.3 | 31.8 | 605 |
| UNEM (CVPR, 2025) | 48.7 | 35.3 | 31.7 | 605 |
| EfficientFSL (AAAI, 2026) | 59.0 | 41.6 | 33.9 | 605 |
| TIM++ (AAAI, 2026) | 57.2 | 40.9 | 34.1 | 605 |
| OSESSL (ICLR, 2026) | 59.1 | 42.5 | 36.9 | 605 |
| DeepDPM (CVPR, 2022) | 44.5 | 38.0 | 40.1 | 775 |
| UNSEEN+DCN (ICDMW, 2024) | 51.1 | 36.0 | 42.3 | 782 |
| SpherePair (NeurIPS, 2025) | 55.2 | 42.4 | 40.0 | 635 |
| SCMax (AAAI, 2026) | 54.7 | 43.3 | 40.6 | 720 |
| DeepDP-TGMM (CVPR, 2026) | 57.6 | 44.6 | 40.8 | 715 |
| DD-FSC (Ours) | 61.2 | 48.2 | 40.2 | 711 |
DD-FSC achieves the highest NMI and ACC and the smallest cluster-count error on Herbarium19. Its ACC exceeds the strongest baseline by 3.6 percentage points; its estimated count is 711, an absolute error of 28. This comparison shows that its gains over baselines extend beyond balanced data to naturally unequal class sizes, unknown category counts, and sparse labels, yielding more accurate overall grouping and count estimation.
We have added the controlled unlabeled class imbalance study in Appendix G.3 (Tables 29 and 30) and the Herbarium19 unknown- few-shot evaluation in Appendix D.4 (Tables 16 and 17), highlighted in blue.
We are truly grateful for the time you have taken to review our paper and your insightful review. Here we address your comments in the following, where W denotes a weakness raised by the reviewer and R denotes our response.
R1-1. Compared with the prior density-based methods you mentioned, DD-FSC’s methodological advance lies in using cluster posteriors to guide density-based refinement and training representations against the refined targets, combining local density with inferred class membership to mitigate inter-class overlap. This distinction concerns how density information is used and how it affects representation optimization. DIPO iteratively refines prototypes toward higher-density regions without additional learnable parameters, but does not update feature representations during prototype refinement; overlap in the representation can therefore persist despite improved prototype locations. The mean-shift propagation (MSP) variant of TAFSSL refines prototypes within a task-adaptive feature subspace without further adjusting that subspace, so residual class overlap can still affect classification. HG-VAE uses high-density hub samples to improve feature generation, but hub selection relies on nearest-neighbor occurrence rather than inter-class separation; improved generation quality therefore does not directly guarantee class separability. CMS combines neighborhood mean shift with contrastive learning to update representations, but its mean-shift step relies on neighborhood aggregation, so cross-class neighbors may interfere with the refinement direction.
We have expanded the discussion of prior density-based methods and their differences from DD-FSC in Appendix E.5, highlighted in blue.
R1-2. We conducted additional controlled comparisons with DIPO-style refinement and nonparametric mean-shift to assess the gains and training cost of learned score guidance. Under the main five-shot, known-K protocol on ImageNet-100, two controls replace only semantic-anchor refinement, retaining score guidance elsewhere. A third replaces all score guidance with mean-shift and removes score-network training. All configurations share data splits, supports, initialization of shared modules, remaining objectives, training epochs, and the final predictor. Following Appendix C.3, timings use a single 12-GiB NVIDIA GeForce RTX 3080 Ti GPU and include score-network training where applicable and training-time DP updates, excluding feature caching and final inference.
DD-FSC improves ACC over DIPO-style and mean-shift anchor refinement by 2.7 and 3.7 percentage points, respectively, with improvements in both NMI and ARI. Compared with replacing all score guidance with mean-shift, DD-FSC improves ACC by 6.3 points and reduces total training time by 65.0% (266.0 to 93.2 minutes). This timing includes score-network training, demonstrating improved clustering with lower training cost in the matched setting.
We have added the controlled comparisons with density-based refinement and the corresponding training-time results in Appendix E.5 (Table 22), highlighted in blue.