- Rating
- 2
- Confidence
- 4
- Informativeness
- 2
- Presentation
- 2
- Soundness
- 2
Read review
Paper summary
This paper proposes OSFGC, a one-shot federated graph classification framework that enables multiple clients to collaboratively train a global graph classifier. The core pipeline is: client trains a class and size conditioned graph diffusion model to generate graphs and constructs a Graph Distribution Sketch that includes class-conditional structural statistics, histograms, and random-feature moments of its local graph data. Both the generator and GDS are uploaded to the server. Then, the server generates synthetic graphs using the uploaded diffusion models, screens candidates based on structural consistency and diversity, and employs CR-KMM to assign weights to the retained samples. Finally, a global classifier is trained on the weighted synthetic dataset, with an additional GDS-based predictive calibration mechanism at inference time.
Critical strengths
- The problem itself is meaningful.
- Promising empirical results across many heterogeneous graph classification settings. The paper evaluates the proposed approach on eight federated graph classification settings, including both single-dataset and cross-dataset scenarios. The reported improvements, particularly in Macro-F1, are encouraging. The authors also provide several component ablations and analyses of structural alignment, which offer preliminary evidence that the proposed mechanisms can improve classification performance.
- The weighting mechanism for each generated is reasonable and is believed to better match the original data distribution.
Critical weaknesses
- The whole pipeline consists of many components and the necessity of some is not clear. For example, the uploaded GDS contains rich class-conditional information, including structural histograms, descriptor means, second moments, and class counts. The predictive calibration mechanism in Eq. (29) already defines a classifier-like scoring function based directly on these statistics. It’s not sure how well a classifier would use only the uploaded GDS perform. Is the other component necessary?
- The paper does not quantify the communication costs of the proposed method. Although OSFGC requires only one communication round, each client must upload a diffusion generator together with a collection of class-conditional distribution statistics. A single communication round does not necessarily imply low communication overhead. And high communication costs would trade the benefit of the one-shot setting.
- The training effectiveness of graph diffusion models under extreme label-distribution skews or situations where each client has very limited local training data is not clear. Especially when the proposed method chooses to train a generator with fine-grained control.
- The performance and robustness of proposed method under privacy protection strategy is not provided. Each client uploads both a locally trained diffusion generator and class-conditional GDS statistics. These statistics contain information about local class distributions, graph sizes, node attributes, connectivity patterns, and other structural properties. Together with the generator parameters, they may expose information about the underlying private training graphs. The Appendix discussed optional Gaussian perturbation and DP-SGD, but how the proposed method perform under such protection is not clear.
- Some one-shot federated learning baselines (e.g., DENSE, Co-Boost) should be included, if applicable. They share a similar paradigm that clients train a data generator and train a classifier on the server.
Questions for authors
- Some implementation choices require clarification. Appendix B.1 describes the server-side classifier as an MLP, whereas several baselines use GIN-based classifiers. The paper does not clearly specify how variable-sized graphs are converted into the MLP input representation.
- How effective is direct classification using only the uploaded GDS (or the variable derived from it)?
- The framework introduces several hyperparameters across candidate screening, distribution matching, and predictive calibration, including , , , , , , and . The selection and sensitiveness of these hyper-parameters need more illustration.
4 comments
Response to Reviewer ppZK (2/4)
In the following, we clarify weaknesses 3 and 4.
3. Generator Utility with Scarce Data and Severe Label Skew. (W3)
Protocol: We retrained local generators after reducing or redistributing PROTEINS training data, retaining the original validation and test partitions. Predictive calibration is disabled, so the classifier uses only the learned predictions from synthetic training graphs. This directly tests whether the generated data retain useful class information. Results from a single run are reported below.
| Training condition | ACC (%) | Macro-F1 (%) |
|---|---|---|
| 100% original training data (reference) | 73.53 | 68.87 |
| Retain 25% of each client's training data | 71.93 | 64.65 |
| Retain 10% of each client's training data | 65.86 | 48.09 |
| Dirichlet label skew (α = 1) | 71.85 | 66.52 |
| Dirichlet label skew (α = 0.1) | 71.78 | 67.83 |
| Dirichlet label skew (α = 0.03) | 68.30 | 52.35 |
| At most 10 graphs per class per client | 71.85 | 66.15 |
| At most 5 graphs per class per client | 71.70 | 69.77 |
| α = 0.1 and retain 25% | 66.47 | 65.42 |
Results: All eight scarcity and label-skew conditions exceed the majority-class ACC reference of 57.8%, demonstrating useful class information in the generated training set under limited local data or severe label imbalance.
Why It Remains Effective: For scarce data: Fine-grained structural control is applied during generation and subsequent candidate screening using empirical histogram guidance, without requiring the denoiser to learn additional complex control parameters during local training. For label skew: Each generator models its locally supported classes, and the server combines their synthetic samples without requiring every client to cover the global label space.
4. Performance under Privacy-oriented Perturbations. (W4)
Protection Tested: We jointly perturb GDS statistics with Gaussian noise and train the diffusion generator using DP-SGD, clipping each graph's gradient to an norm of at most before adding Gaussian noise with standard deviation , where is the noise multiplier. GDS perturbations cover histograms, counts, moments, and reliability scores. We add Gaussian noise independently to each component according to its upload scale. We use , progressively increasing the noise strength. The differential privacy budget quantifies privacy protection: smaller values provide stronger protection against inferring from the released generator parameters whether an individual graph participated in training. The reported generator-only estimates decrease from 10.0 to 4.4 (, with neighboring datasets differing by the addition or removal of one graph), within the approximately 1–10 range commonly studied in DP-SGD research. Values below are three-seed mean ACC / Macro-F1 (%); the no-noise reference is the original Table 1 result.
| Generator σ | Generator ε (reported estimate) | PROTEINS | BIO-SN-CV |
|---|---|---|---|
| 0 (no-noise reference) | — | 76.68 / 73.36 | 49.97 / 42.88 |
| 0.50 | 10.0 | 76.43 / 73.06 | 49.78 / 42.30 |
| 0.60 | 8.0 | 75.91 / 72.39 | 47.38 / 39.82 |
| 0.75 | 6.1 | 75.96 / 71.95 | 46.70 / 38.94 |
| 1.00 | 4.4 | 75.16 / 70.27 | 44.59 / 36.60 |
Results: The method retains substantial predictive utility under simultaneous perturbations of both uploaded information sources, with greater degradation under stronger noise and cross-domain heterogeneity.
Response to Reviewer ppZK (3/4)
In the following, we address weakness 5 on the additional one-shot baselines.
5. Additional DENSE and Co-Boost Baselines. (W5)
We added DENSE and Co-Boost with GIN teachers trained on local original graphs and uploaded once, followed by server-side soft-graph generation and GIN student training, using the same GIN backbone as FedAvg, FedProx, and GCFL+ to accommodate graph data. Their core algorithms directly call the official implementations, and training hyperparameters follow the official configurations. The following results are mean ± standard deviation (%), with OSFGC results taken directly from the original Table 1.
| Setting | DENSE (ACC / Macro-F1) | Co-Boost (ACC / Macro-F1) | OSFGC (ACC / Macro-F1) |
|---|---|---|---|
| PTC_MR | 57.07 ± 0.00 / 36.33 ± 0.00 | 52.78 ± 4.29 / 34.48 ± 1.84 | 68.18 ± 0.32 / 64.76 ± 0.25 |
| NCI1 | 59.16 ± 6.31 / 48.73 ± 12.47 | 65.13 ± 1.38 / 64.01 ± 1.63 | 67.37 ± 0.18 / 66.45 ± 0.15 |
| PROTEINS | 63.28 ± 4.19 / 45.78 ± 5.57 | 71.55 ± 2.83 / 64.84 ± 0.30 | 76.68 ± 0.01 / 73.36 ± 0.02 |
| DD | 50.00 ± 10.57 / 32.94 ± 4.73 | 61.56 ± 0.99 / 45.75 ± 8.07 | 65.58 ± 0.23 / 56.99 ± 0.20 |
| CHEM | 54.53 ± 3.07 / 40.32 ± 2.93 | 55.60 ± 1.22 / 42.07 ± 1.19 | 74.44 ± 0.10 / 65.24 ± 0.05 |
| BIO-SN-CV | 24.14 ± 0.00 / 13.81 ± 0.00 | 23.20 ± 9.05 / 12.12 ± 5.03 | 49.97 ± 0.35 / 42.88 ± 0.35 |
| BIO-CHEM | 54.62 ± 0.22 / 42.12 ± 0.59 | 51.92 ± 4.08 / 38.17 ± 4.96 | 66.11 ± 0.11 / 59.12 ± 0.22 |
| BIO-CHEM-SN-CV | 34.81 ± 0.00 / 21.50 ± 0.00 | 41.15 ± 0.48 / 25.84 ± 1.84 | 56.24 ± 2.02 / 50.13 ± 0.77 |
Results: OSFGC achieves higher mean ACC and Macro-F1 than both DENSE and Co-Boost in all eight settings.
Response to Reviewer ppZK (4/4)
In the following, we address questions 1 and 3 on graph inputs and hyperparameters.
6. Server MLP and Variable-size Graph Inputs. (Q1)
Input Representation: Each graph, whether synthetic or a real test graph, is first mapped by the same deterministic function to a fixed-length graph-level descriptor. This vector consists of (i) bounded structural statistics, including normalized graph size, edge density, degree statistics, clustering, connectivity, triangle density, and low-order spectral moments of the normalized Laplacian; and (ii) channel-wise means and standard deviations of the selected node features, with node features first zero-padded to a common width across datasets. Graphs of different sizes therefore receive the same form of representation before MLP classification.
Design Rationale: (i) The server trains not on a massive collection of real graphs, but on a limited number of screened synthetic graphs generated for each eligible class by each client's generator and assigned fractional weights by CR-KMM, while cross-domain zero-padding brings node-feature dimensionality to nearly 500. Training a deep message-passing GNN from scratch on this limited synthetic pool could overfit sampling artifacts, including discrete edge perturbations. Mapping these graphs to bounded graph-level statistics consistent with GDS and training a lightweight MLP is therefore intended to limit overfitting. (ii) The w/o DG ablation uses exactly the same descriptor inputs and MLP architecture and retains GDS predictive calibration, but replaces conditional diffusion with GDS-based random graph sampling. As shown in Figure 5 of the main paper, retaining the same MLP architecture while removing only diffusion reduces PROTEINS ACC from 76.68% to 62.75% and Macro-F1 from 73.36% to 39.41%, while BIO-SN-CV ACC falls from 49.97% to 30.63%. Because the representation and classifier are held fixed, these gains cannot be attributed to an advantage from the descriptor inputs or MLP architecture. Instead, they support the contribution of conditional graph diffusion to the utility of the synthetic graphs, consistent with better preservation of joint structural and attribute information.
7. Hyperparameter Configuration and Sensitivity. (Q3)
Thank you for the valuable comment. Across all eight benchmarks, we fix the candidate screening and CR-KMM weights without per-dataset tuning: , , , , , and . The default settings are and . We use in the main experiments. Figure 6 reports sensitivity over . A value of does not by itself mean that predictions rely on GDS alone: (i) in prototype calibration (Appendix A.5), scales standardized GDS scores before they are added to unstandardized classifier logits, so its effect depends on the score differences between competing classes. (ii) In cross-domain calibration, both the classifier and GDS scores are standardized. Very low GDS scores can suppress implausible out-of-domain classes while leaving the classifier to distinguish among plausible alternatives. At a fixed variance floor, Figure 6 shows little change in ACC or Macro-F1 over . (iii) Full OSFGC outperforms GDS-only in both metrics on both datasets (Table 2). Even with calibration disabled (), Classifier-only achieves higher mean ACC and Macro-F1 on both PROTEINS and BIO-SN-CV (Table 2) than the four global shared-head one-shot baselines DENSE, Co-Boost, GHOST, and OASIS (Table 1).
Finally, thanks again for your time and helpful comments. We hope your concerns are well addressed. For any other questions, we are happy to discuss them and make further clarifications.
Response to Reviewer ppZK (1/4)
We thank the reviewer for the valuable and constructive comments. We provide direct component comparisons, communication measurements, stress tests, privacy-noise experiments, and additional one-shot baselines below.
1. Is Classification from GDS Alone Sufficient? (W1, Q2)
Direct Comparison: We evaluated three prediction rules: GDS-only, classifying solely from the uploaded GDS; Classifier-only, using the classifier trained on CR-KMM-weighted synthetic graphs with predictive calibration disabled; and Full OSFGC, combining both. Mean ACC and Macro-F1 in percent over three repeated runs are reported below.
Results:
w/o DG(62.75 / 39.41 and 30.63 / 17.36),w/o D-GDS-G(72.19 / 65.11 and 33.63 / 18.21),w/o CS(73.78 / 70.61 and 36.61 / 27.13), andw/o CR-KMM(73.30 / 67.85 and 34.22 / 21.64). These results show that each stage contributes to performance in the tested pipeline.2. Measured Communication Payload. (W2)
We measured OSFGC's complete upload files, including generator parameters, GDS, and metadata. For CeFGC and OASIS, we count float32 parameter bytes in the uploaded objects received by their official server implementations: CeFGC uploads one label-conditioned EDP-GNN diffusion generator per client, using its communication-saving variant, while each OASIS client uploads a synthesizer and a teacher model with a codebook. All values below are summed across clients and reported in MiB. The OSFGC generator and GDS columns count float32 tensor bytes. Metadata and file-format overhead is estimated as the total upload minus the generator and GDS values, using the rounded figures shown.
Results: OSFGC has a smaller upload payload than CeFGC and OASIS in all eight settings. GDS accounts for only 0.02–0.38 MiB across all clients, approximately 3%–7% of OSFGC's uploaded tensor bytes, with generator parameters accounting for the remainder. The gap to CeFGC is largest in cross-domain settings, where its generator size grows substantially with node-feature dimensionality.