Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Description
本 PR 根据 ICDE 2025 论文 Effective and Scalable Heterogeneous Graph Neural Network Framework with Convolution-oriented Attention,为 OpenHGNN 增加 V-HCAN 和 D-HCAN。
Changes
HCAN(V-HCAN),对应论文算法 1 和公式 (4)-(11):1 - gamma;DHCAN/D-HCAN,对应论文公式 (13)-(15):N_i*;Micro_f1等指标早停,V-HCAN 不再按损失选择 checkpoint。Formula Correspondence
Verification
22 passed:python -m compileall通过。git diff --cached --check通过。最终公式版 V-HCAN 在 HGBn-ACM 单次验证中得到 Macro-F1 96.30%、Micro-F1 96.13%,训练和早停流程正常。
D-HCAN 已在完整 OGB-MAG 上完成单次大规模训练验证;
batch_size=10000时平均训练时间约 6.30 秒/轮,与论文表 VIII 的 5.89 秒/轮接近。Reproduction Note
论文没有发布官方代码,也没有完整给出 D-HCAN 的隐藏维度、前馈网络深度、dropout、batch size,以及 OGB-MAG 无特征节点和反向关系的具体预处理方式。因此,本 PR 以论文公开的数学定义为实现依据,并保证公式级对应;对于论文未公开的工程细节采用 OpenHGNN 的现有数据处理方式和显式配置项,不声明默认配置可以逐项复现表 VIII 的五次运行均值。
在当前 OpenHGNN 预处理和已测试默认结构下,D-HCAN 的单次 OGB-MAG 准确率未达到论文表 VIII。该差异在 PR 中明确披露,不通过标签传播、多阶段训练或额外嵌入等论文未说明的技巧追数值。
Checklist
[Model]开头