- Rating
- 2
- Confidence
- 4
- Informativeness
- 2
- Presentation
- 2
- Soundness
- 2
Read review
Paper summary
This paper proposes HODA for open-set graph domain adaptation under heterophily. The method combines multiscale low-pass and high-pass filtering with prototype adaptation and classifier–clustering consensus to classify known target nodes and reject unknown ones. Experiments evaluate the framework on Blog, WebKB, and MAG.
Critical strengths
-
The experiments include heterophilic GNNs as well as closed-set and open-set GDA baselines. HODA improves performance on several reported transfer tasks, and the ablations help identify the effects of high-pass filtering, prototype adaptation, and consensus selection.
-
Correcting the inconsistent class-label encodings across WebKB graphs is useful. The authors recover the original semantic labels and establish a consistent mapping across universities, addressing a concrete issue that could otherwise confound cross-domain comparisons.
Critical weaknesses
-
Restrictive feature-space assumptions. The source and target graphs share the same feature dimension and use shared input MLP. This leaves feature heterogeneity, an important challenge in cross-domain graph learning [1], outside the evaluation. It remains unclear whether HODA can distinguish known classes affected by changes in feature semantics from genuinely unknown classes. An evaluation on different feature spaces or shared language-model encodings of raw text would strengthen its practical relevance.
-
Limited architectural novelty. Low-pass, high-pass filtering and identity channels are already well established in heterophilic graph learning [2]. I do not see a clear architectural advance in applying these filters at multiple propagation orders and combining their outputs with MLPs. The paper needs to establish what this design contributes beyond an existing multi-pass graph encoder in heterophilic graphs combined with open-set adaptation.
-
Insufficient evaluation on larger heterophilic graphs. WebKB contains only 183–251 nodes per graph, and Blog contains roughly 2,300–2,900. MAG adds scale, but the paper does not report homophily statistics for its evaluated splits. Larger established heterophilic benchmarks, such as arXiv-year and snap-patents [3], could support more demanding transfer evaluations. Runtime and memory measurements are also missing, leaving scalability unclear.
-
Limited applicability of the theoretical analysis. Theorem 2 assumes a regular graph, binary classes, identical local homophily at every node, and restrictions on the interference spectrum. These conditions are quite restrictive. Real graphs can contain neighborhoods with very different homophily levels, for which different filters may be appropriate. The analysis provides limited guidance for this node-level variation or for multiclass problems. I would find an explanation of filter suitability across different neighborhoods more useful. Theorem 1's observation that additional views cannot increase the ideal joint risk is also a straightforward consequence of the nested hypothesis-class assumption.
References: [1] Liu, H., et al. “One for All: Towards Training One Graph Model for All Classification Tasks.” International Conference on Learning Representations (ICLR), 2024.
[2] Luan, S., et al. “Revisiting Heterophily for Graph Neural Networks.” Advances in Neural Information Processing Systems (NeurIPS), 2022.
[3] Lim, D., et al. “Large Scale Learning on Non-Homophilous Graphs: New Benchmarks and Strong Simple Methods.” Advances in Neural Information Processing Systems (NeurIPS), 2021.
Questions for authors
-
How would HODA handle different source and target feature dimensions or semantics?
-
What does the proposed encoder offer beyond existing heterophily-aware architectures?
-
How does HODA perform on larger heterophilic graphs, and what are its runtime and memory costs?
7 comments
Official Comment by Authors (2/5)
To examine what our design contributes beyond an existing multi-pass encoder combined with open-set adaptation, we replace each part of HODA with its standard counterpart, one at a time: (1) ACM encoder replaces our multiscale low- and high-pass encoder with a tuned ACM-GCN; (2) Standard DEC replaces the reliability-weighted mixing of source and target prototypes with freely learned DEC centers; (3) MSP / Entropy / Energy replace classifier-prototype consensus with confidence-based selection. The results (ACC / HS, %) are summarized below.
| Variant | Blog | WebKB | Citation |
|---|---|---|---|
| HODA | 59.64 / 61.21 | 80.28 / 76.98 | 66.31 / 67.72 |
| ACM encoder | 57.06 / 57.65 | 70.55 / 63.89 | 64.58 / 67.86 |
| Standard DEC | 58.45 / 60.06 | 75.99 / 63.90 | 53.78 / 54.01 |
| MSP | 58.74 / 60.16 | 75.41 / 72.58 | 66.39 / 63.58 |
| Entropy | 58.17 / 59.67 | 76.27 / 72.80 | 65.47 / 60.38 |
| Energy | 58.05 / 59.13 | 74.80 / 63.86 | 64.14 / 64.92 |
None of the standard counterparts is reliable across benchmarks. The ACM encoder matches ours on the homophilic Citation benchmark but falls behind under heterophily, especially on WebKB. Standard DEC is comparable on Blog but drops substantially on WebKB and Citation. Confidence-based selection yields lower HS than consensus on all three benchmarks. These results show that an existing encoder and standard open-set components cannot be combined directly in the joint setting.
To understand these failures, we examine which target nodes each variant selects on WebKB, using the quantities in Theorem 1. Here, is the fraction of known-class nodes wrongly excluded from alignment, is the fraction of unknown nodes among those retained, and Precision is the accuracy of the known-class pseudo labels. All values are averaged over five seeds and six transfer tasks.
| Variant | Precision | HS | ||
|---|---|---|---|---|
| HODA | 0.239 | 0.213 | 0.777 | 76.98 |
| ACM encoder | 0.228 | 0.395 | 0.548 | 63.89 |
| Standard DEC | 0.471 | 0.157 | 0.870 | 63.90 |
| MSP | 0.269 | 0.264 | 0.702 | 72.58 |
| Entropy | 0.279 | 0.242 | 0.720 | 72.80 |
| Energy | 0.472 | 0.173 | 0.830 | 63.86 |
With the ACM encoder, far more unknown nodes are retained and the pseudo labels are much less accurate, which indicates that known and unknown nodes are poorly separated in its representations. Unlike ACM, our encoder uses parameter-free filters at multiple propagation orders and keeps the channels separate. The other variants fail for different reasons. Standard DEC and energy-based selection exclude nearly half of the known-class nodes, while MSP and entropy selection exclude more known nodes and retain more unknown ones than consensus. These results show that combining an existing heterophily-aware encoder with open-set adaptation is not sufficient. Our encoder does not introduce a new filter. Its contribution is to provide representations on which prototype adaptation and consensus selection remain reliable under heterophily. We will add these results to the revised version and clarify our contributions accordingly.
Official Comment by Authors (3/5)
3. Insufficient evaluation on larger heterophilic graphs. (W3/Q3)
Response: Thank you for these suggestions. We address the three points in turn.
Homophily. The homophily, the class counts after the split, and the share of unknown nodes are listed below for all graphs. Homophily is computed on the original classes.
| Benchmark | Graph | Edge homophily | Node homophily | Nodes per known class | Nodes per unknown class | Share of unknown nodes |
|---|---|---|---|---|---|---|
| WebKB | Cornell | 0.123 | 0.111 | 38 / 16 / 30 | 82 / 17 | 54.1% |
| Texas | 0.061 | 0.057 | 33 / 1 / 30 | 101 / 18 | 65.0% | |
| Wisconsin | 0.178 | 0.155 | 70 / 10 / 32 | 118 / 21 | 55.4% | |
| Blog | Blog1 | 0.399 | 0.389 | 399 / 320 / 419 / 392 | 377 / 393 | 33.5% |
| Blog2 | 0.400 | 0.390 | 486 / 457 / 502 / 466 | 515 / 470 | 34.0% | |
| Citation | ACMv9 | 0.800 | 0.794 | 1916 / 2767 / 2110 | 803 / 1764 | 27.4% |
| Citationv1 | 0.860 | 0.832 | 2262 / 2323 / 2004 | 690 / 1656 | 26.3% | |
| DBLPv7 | 0.820 | 0.806 | 1188 / 1808 / 1307 | 332 / 849 | 21.5% |
For MAG, the numbers of known and unknown nodes under the two open-set settings are as follows.
| Graph | Edge homophily | Node homophily | Known / unknown nodes (1 unknown class) | Share of unknown nodes | Known / unknown nodes (5 unknown classes) | Share of unknown nodes |
|---|---|---|---|---|---|---|
| US | 0.516 | 0.549 | 70,709 / 61,849 | 46.7% | 65,654 / 66,904 | 50.5% |
| CN | 0.530 | 0.587 | 38,021 / 63,931 | 62.7% | 31,628 / 70,324 | 69.0% |
| DE | 0.529 | 0.553 | 21,614 / 21,418 | 49.8% | 19,567 / 23,465 | 54.5% |
| JP | 0.563 | 0.595 | 14,957 / 22,541 | 60.1% | 13,329 / 24,169 | 64.5% |
| RU | 0.768 | 0.803 | 4,874 / 27,959 | 85.2% | 4,288 / 28,545 | 86.9% |
| FR | 0.569 | 0.573 | 13,917 / 15,345 | 52.4% | 12,886 / 16,376 | 56.0% |
arXiv-year and snap-patents. These are valuable benchmarks for heterophilic learning on a single graph. However, they are not directly suitable for domain adaptation, which requires separate source and target graphs that share the same classes. Our benchmarks instead follow those commonly used in graph domain adaptation under heterophily and in open-set graph domain adaptation: Blog and Citation [1, 2], WebKB [2, 3], and MAG [1, 4]. To our knowledge, none of these works evaluates on arXiv-year or snap-patents. Among these benchmarks, MAG provides the scale, where the largest graph has more than 130,000 nodes and nearly 700,000 edges.
Runtime and memory. We first analyze the complexity of HODA. The multiscale filters require sparse matrix multiplications per graph, which cost and are computed once before training. During training, the encoder applies MLPs to the precomputed views without further message passing, so each epoch costs , where is the hidden dimension. Prototype computation and the DEC loss cost , and the alignment loss is computed on at most 512 sampled nodes per domain, independent of the graph size. The per-epoch cost of HODA is therefore linear in the number of nodes, and the dependence on the number of edges appears only in the one-time preprocessing.
We also measure the training time and the peak GPU memory of all methods on one NVIDIA L4 GPU with their reported configurations. The results on B1$\rightarrow$B2 and on the largest MAG task, US$\rightarrow$CN, are listed below.
| Category | Method | B1$\rightarrow$B2 time (s) | B1$\rightarrow$B2 memory (GB) | US$\rightarrow$CN time (s) | US$\rightarrow$CN memory (GB) |
|---|---|---|---|---|---|
| Heterophilic GNNs | ACM | 2.0 | 0.03 | 8.9 | 0.3 |
| CMGNN | 1.4 | 0.18 | 105 | 1.3 | |
| HeroFilter | 1.8 | 0.05 | 187 | 0.9 | |
| H2GCN | 10.5 | 0.77 | 93 | 9.8 | |
| HIMP | 3.7 | 0.12 | 79 | 2.7 | |
| Closed-set DA | ADAlign | 4.6 | 0.6 | 222 | 6.2 |
| DiffGDA | 288 | 12.6 | 623 | 13.5 | |
| Pair-Align | 96 | 0.24 | 1173 | 13.4 | |
| DGSDA | 123 | 4.4 | 364 | 7.7 | |
| DFT | 79 | 1.1 | 789 | 8.7 | |
| RSGDA | 1.9 | 0.06 | 19 | 1.2 | |
| Open-set DA | UAGA | 18.6 | 0.08 | 798 | 2.0 |
| GraphRTA | 9.9 | 1.1 | 43 | 1.5 | |
| Ours | HODA | 3.5 | 0.12 | 68 | 6.0 |
H2GCN is implemented in TensorFlow, and its memory is reported by the TensorFlow allocator. HODA is 3 to 17 times faster than the closed-set and open-set DA methods on US$\rightarrow$CN, except RSGDA and GraphRTA, and its memory is comparable to or lower than that of the alignment-based methods. Its cost is of the same order as that of the heterophilic GNNs, which perform no adaptation, and it is faster than H2GCN, the strongest baseline. The multiscale filters take less than 0.1 seconds on every task. Across all 34 tasks, HODA trains in under 90 seconds and uses at most 6.5 GB of memory. We will add the complexity analysis and these measurements to the revised version.
Official Comment by Authors (1/5)
We sincerely thank you for your detailed and rigorous review. We have addressed each of them below.
1. Restrictive feature-space assumptions. (W1/Q1)
Response: Thank you for raising this point. We agree that feature heterogeneity is an important challenge in cross-domain graph learning. However, it is orthogonal to the problem studied in this paper. We follow the standard setting of graph domain adaptation, where the source and target graphs share a feature space, and all baselines and benchmarks in our evaluation adopt the same setting. Our focus is on the difficulty that arises within this setting when structural heterophily and target-private classes occur together. HODA does not address different feature dimensions or semantics by itself. In such cases, the two graphs would first need to be mapped into a common input space, after which HODA can be applied without modification.
Within a shared feature space, the concern that known classes affected by feature shift may be confused with genuinely unknown classes is the ambiguity that HODA is designed to resolve. Both kinds of nodes can receive low confidence, so HODA does not rely on confidence alone. It adapts each class prototype toward the target distribution according to its reliability and selects a node only when the classifier and the prototypes agree. We will discuss feature heterogeneity as future work in the revised version.
2. Limited architectural novelty. (W2/Q2)
Response: Thank you for this comment. We agree that low-pass, high-pass, and identity channels are well established in heterophilic graph learning, and we do not claim the encoder as a new architecture. However, these encoders were developed for a single labeled graph, and combining them with open-set adaptation is non-trivial. Under heterophily, an uncertain target prediction may come from a poorly transferred known-class node or from a genuinely unseen class, and existing methods cannot tell the two apart. Heterophilic GNNs neither align domains nor recognize unknown classes. Closed-set DA methods wrongly align unknown nodes to known classes. Open-set DA methods reject low-confidence nodes as unknown, although many known nodes also have low confidence under heterophily. The results below report ACC / HS (%) averaged over the transfer tasks of each benchmark (Tables 1, 2, 6 and 7 of the paper), with MAG averaged over both unknown-class settings.
Existing heterophily-aware encoders are the strongest baselines, but they still trail HODA by 2.70 to 4.96 points in HS, since they handle neither domain shift nor unknown classes. Closed-set DA methods are less reliable under heterophily, and even the heterophily-aware RSGDA performs poorly on MAG. Open-set DA methods perform well on the homophilic Citation benchmark but drop sharply on Blog and WebKB.