- Rating
- 1
- Confidence
- 4
- Informativeness
- 3
- Presentation
- 3
- Soundness
- 3
Read review
Paper summary
This paper proposes CGBST, which models the local region structure through Granular Ball and establishes the corresponding relationship of regions by sharing samples across views, thereby restoring the sample relationships of missing views. The experiments demonstrate a certain improvement in clustering performance, but the paper still has significant deficiencies in terms of research motivation, method innovation, and experimental persuasiveness.
Critical strengths
- The overall structure of the paper is clear, and the description of the methods is relatively complete.
- Good clustering results were achieved on multiple datasets, and certain ablation analyses were provided.
Critical weaknesses
- The research significance is limited. The core motivation of the paper is to reduce the sensitivity of instance-level relationships to noise through regional-level structure transmission. However, it does not adequately explain how serious this issue is in the existing IMVC methods, nor does it provide direct evidence to prove that Granular Ball is the necessary choice to solve this problem.
- The method lacks innovation. The overall framework mainly combines regional division, cross-view correspondence based on shared samples, graph relationship restoration, as well as conventional contrastive learning and graph regularization. The combination of these modules is somewhat reasonable, but it lacks sufficient prominent new ideas or theoretical insights.
Questions for authors
- Can the author provide more direct empirical evidence to demonstrate that the existing instance-level structure transfer methods do indeed have significant flaws in noisy or cross-view overlapping sparse scenarios? It is currently unclear how important the problems solved by introducing Granular Ball are.
- Table 2 shows that the Instance-level Transfer outperforms the full method in some metrics on the CUB dataset. Can the author further explain the core advantages of Granular Ball compared to instance-level transfer and prove the necessity of this design by comparing it with other region aggregation or prototype methods?
- The missing rate defined in the paper is η/V, and the actual missing rate is relatively limited. Can the author provide experimental results with a higher actual missing rate, especially in cases where cross-view shared samples are very few, to verify the effectiveness of the method in the claimed difficult scenarios?
4 comments
R2. Methodological Novelty and Advantages over Instance-Level and Prototype-Based Transfer (W2 & Q2)
The central contribution of CGBST is a cross-view structural transfer mechanism that uses granular balls as intermediate units for missing-relation recovery. Existing granular-ball-based multi-view clustering methods primarily employ regional structures for representation learning or multi-granularity alignment. CGBST extends granular-ball modeling to incomplete-view structure recovery through three connected mechanisms: adaptive view-specific granular regions, co-observed-anchor-driven soft region correspondences, and target-topology-guided sample relation recovery. This establishes a complete transfer pathway from source-view regional organization to missing sample relations in the target view.
To further evaluate the contribution of granular-ball modeling, we have added two complementary controls: Instance-level comparison (Page 8, Table 2). The original Instance-level Transfer retains granular-ball construction and replaces region-level correspondence with instance-anchor transfer. We additionally introduce Instance-pure Transfer, which removes granular-ball construction and directly transfers instance-level graph relations. Overall, CGBST outperforms Instance-pure Transfer in 13 out of 16 evaluation metrics across all four datasets, demonstrating the effectiveness of granular-ball structure modeling for cross-view relation recovery.
Prototype replacement (Page 19, Appendix F.3, Table 9). We replace adaptive granular-ball construction with fixed-K K-means prototypes , keeping the remaining structure-transfer and training pipeline unchanged. CGBST outperforms the K-means prototype control across all four metrics at every evaluated missing rate on MSRCV1. At , the improvements reach 6.11, 5.74, 9.73, and 7.14 percentage points in ACC, NMI, ARI, and F-score, respectively.
Together, these comparisons demonstrate the contribution of adaptive granular regions and region-mediated correspondences to recovering missing neighborhood structure.
Table R2(a). Instance-level transfer ablation at (Mean, %; extracted from Page 8, Table 2).
| Dataset | Variant | ACC | NMI | ARI | F-score |
|---|---|---|---|---|---|
| CUB | Full CGBST | 77.37 | 76.64 | 65.51 | 73.01 |
| Instance-pure | 77.94 | 75.64 | 65.06 | 70.87 | |
| MSRCV1 | Full CGBST | 93.40 | 85.93 | 84.40 | 87.32 |
| Instance-pure | 92.11 | 84.51 | 82.05 | 85.85 | |
| OutdoorScene | Full CGBST | 74.13 | 60.79 | 52.40 | 60.50 |
| Instance-pure | 71.14 | 58.42 | 50.06 | 58.29 | |
| NUS-WIDE | Full CGBST | 55.04 | 53.08 | 34.53 | 52.29 |
| Instance-pure | 54.57 | 53.95 | 34.39 | 52.32 |
Table R2(b). K-means prototype replacement on MSRCV1 (Mean, %; Page 19, Appendix F.3, Table 9).
| Variant | ACC | NMI | ARI | F-score | |
|---|---|---|---|---|---|
| 0.1 | K-means prototype (K=20) | 92.10 | 84.69 | 81.87 | 85.87 |
| CGBST (ours) | 94.68 | 88.11 | 86.30 | 89.51 | |
| 0.3 | K-means prototype (K=20) | 90.05 | 82.94 | 78.69 | 83.52 |
| CGBST (ours) | 93.40 | 85.93 | 84.40 | 87.32 | |
| 0.5 | K-means prototype (K=20) | 85.49 | 77.07 | 71.19 | 77.46 |
| CGBST (ours) | 91.60 | 82.81 | 80.92 | 84.60 | |
| 0.7 | K-means prototype (K=20) | 82.76 | 75.61 | 68.54 | 75.14 |
| CGBST (ours) | 86.05 | 78.23 | 73.87 | 78.88 |
R3. Evaluation under Higher Missing Rates and Sparse Co-Observed Samples (Q3)
We have extended the missing-rate evaluation to (Page 18, Appendix F.1, Table 7) and additionally evaluated sparse cross-view overlap through directly controlled common observation rates (Page 9, Figure 3; Page 18, Appendix F.2, Table 8). At , CGBST achieves the best ACC, NMI, and ARI on both MSRCV1 and OutdoorScene among the compared methods, obtaining 86.05% ACC, 78.23% NMI, and 73.87% ARI on MSRCV1, and 72.03% ACC, 57.71% NMI, and 48.99% ARI on OutdoorScene.
The partial-overlap experiments further evaluate scenarios with limited co-observed anchors. At the lowest target common observation rate , the realized pairwise co-observation rates are 28.10–34.29% on MSRCV1 and 28.61–32.03% on OutdoorScene. CGBST achieves higher performance than pure instance-level transfer across all four metrics on both datasets.
These complementary experiments demonstrate the advantages of CGBST over instance-level transfer in recovering neighborhood structure under high missing rates and sparse cross-view observations.
Table R3. Supplementary comparison at (Mean, %; Page 18, Appendix F.1, Table 7).
| Method | MSRCV1 ACC | NMI | ARI | F-score | OutdoorScene ACC | NMI | ARI | F-score |
|---|---|---|---|---|---|---|---|---|
| Mask-IMvC | 83.06 | 75.26 | 66.85 | 75.02 | 55.80 | 45.99 | 32.06 | 45.14 |
| DSIMVC++ | 35.24 | 26.29 | 15.80 | 30.87 | 55.58 | 45.55 | 32.92 | 55.38 |
| GIGA | 39.52 | 30.75 | 12.76 | 38.74 | 20.22 | 8.55 | 0.98 | 13.04 |
| BURG | 69.21 | 61.05 | 51.66 | 65.29 | 41.83 | 36.68 | 21.74 | 40.20 |
| CPSPAN | 70.00 | 58.74 | 50.41 | 67.53 | 62.69 | 50.51 | 40.36 | 62.70 |
| CPSPAN-mask | 74.60 | 65.53 | 57.24 | 72.65 | 64.17 | 51.90 | 42.13 | 64.19 |
| DCCA | 35.71 | 22.91 | 10.30 | 37.56 | 21.59 | 7.26 | 2.68 | 22.22 |
| DCG | 50.79 | 31.06 | 22.53 | 50.17 | 66.67 | 56.57 | 47.47 | 66.50 |
| GFSAF | 20.63 | 4.61 | -0.38 | 19.84 | 15.30 | 0.46 | 0.04 | 14.38 |
| SMHGC | 27.62 | 12.32 | 5.11 | 24.91 | 21.07 | 7.59 | 3.79 | 19.59 |
| DIMVC-HIA | 19.52 | 6.11 | 2.30 | 9.05 | 60.73 | 48.99 | 37.46 | 60.76 |
| SCM | 78.73 | 70.81 | 60.63 | 79.15 | 66.82 | 54.89 | 44.81 | 66.64 |
| CGBST (ours) | 86.05 | 78.23 | 73.87 | 78.88 | 72.03 | 57.71 | 48.99 | 57.88 |
We would like to follow up on our responses and the additional analyses presented in the rebuttal. We hope that the new experimental evidence and clarifications have addressed your questions regarding the methodological contributions and empirical evaluation of CGBST. We welcome any further feedback and would be happy to clarify any remaining concerns.
R1. Empirical Evidence for Granular-Ball Transfer under Noise and Sparse Cross-View Overlap (W1 & Q1)
We appreciate the reviewer's suggestion to further evaluate CGBST under sparse cross-view overlap and noisy conditions. We have added a dedicated evaluation comparing CGBST with pure instance-level transfer under noise perturbations and partial cross-view overlap (Page 9, Figure 3; Page 18, Appendix F.2, Table 8). The experiments vary the common observation rate ρ∈{0.3,0.4,0.5} at a fixed noise level σ=0.25, with the actual pairwise co-observation rates explicitly reported.
CGBST consistently outperforms pure instance-level transfer across all four metrics on MSRCV1, achieving improvements of 7.15, 6.16, 8.88, and 5.16 percentage points in ACC, NMI, ARI, and F-score at ρ=0.3, respectively. It also improves all four metrics on OutdoorScene at ρ=0.3 and ρ=0.4. These results demonstrate the advantages of region-mediated structure transfer under sparse and noisy cross-view observations.
Table R1. Partial cross-view overlap and noise robustness (σ=0.25; Mean, %).
Pairwise co-observation rates for ρ=0.3/0.4/0.5: MSRCV1 28.10–34.29% / 28.57–42.86% / 45.71–52.86%; OutdoorScene 28.61–32.03% / 38.06–40.36% / 49.40–51.53%.