发表机构
University of Catania; University of Cambridge; University Campus Bio-Medico of Rome（卡塔尼亚大学; 剑桥大学; 罗马生物医学自由大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
ProtoGuide提出一种事后、与骨干无关的框架，通过原型评分和梯度注入实现离散图扩散模型的类条件引导，在多个数据集上显著提升分类准确率。
AI 中文摘要
离散扩散模型是图生成领域的一个重要家族，但标准的类条件机制在训练期间将类信号嵌入去噪器中，从而将条件机制与训练后的模型绑定在一起。分类器引导通过在连续域中利用分类器的梯度来引导冻结模型，从而避免了这种耦合，但离散图扩散模型采样的是离散的边状态，因此梯度无法通过采样的图进行传播。我们引入了ProtoGuide，这是一种事后（post-hoc）、与骨干网络无关的框架，它恢复了类似的机制。在每个反向步骤中，去噪器对每条边的输出被松弛为可微的软邻接矩阵，由冻结的孪生图神经网络嵌入，并与目标类原型及其最近的竞争对手进行评分；由此产生的每条边的梯度，通过余弦调度进行衰减，被注入回去噪器的输出中。所有组件保持冻结，因此通过提供不同的原型即可重新定向引导。在五类真实世界网络和两个架构不同的骨干网络EDGE和DiGress上，ProtoGuide将宏分类准确率从50.7%提升至73.5%，以及从73.6%提升至83.8%，并在我们的配置下优于DiGress的内置条件训练。在未引导模型最弱的类别上，提升幅度最大，且提升并非在所有类别上均匀分布。在大多数设置中，每图覆盖率保持较高，而分布效应则依赖于类别。一个Best-of-N选择基线在给定足够过采样的情况下可以达到相同的准确率，但会显著牺牲图的多样性。一个独立初始化的分类器、一个方向性测试以及一个少样本分析支持目标定向引导和对极小支持集的鲁棒性。
英文摘要
Discrete diffusion models are a prominent family for graph generation, but standard class-conditional mechanisms embed the class signal in the denoiser during training, tying the conditioning mechanism to the trained model. Classifier guidance avoids this coupling in continuous domains by steering a frozen model with a classifier's gradient, but discrete graph diffusion samples discrete edge states, so gradients cannot propagate through the sampled graph. We introduce ProtoGuide, a post-hoc, backbone-agnostic framework that recovers an analogous mechanism. At each reverse step the denoiser's per-edge output is relaxed into a differentiable soft adjacency, embedded by a frozen Siamese graph neural network, and scored against a target-class prototype and its nearest competitor; the resulting per-edge gradient, damped by a cosine schedule, is injected back into the denoiser output. All components stay frozen, so guidance is retargeted by supplying a different prototype. On five classes of real-world networks and two architecturally different backbones, EDGE and DiGress, ProtoGuide raises macro classification accuracy from 50.7% to 73.5% and from 73.6% to 83.8%, and outperforms DiGress's built-in conditional training under our configuration. Gains are largest where the unguided models are weakest, and are not uniform across classes. Per-graph coverage remains high in most settings, while distributional effects are class-dependent. A Best-of-N selection baseline matches this accuracy given enough oversampling, but at a substantial cost in graph diversity. An independently initialized classifier, a directionality test, and a few-shot analysis support target-directed steering and robustness to very small support sets.
CommentsPreprint. Submitted to IEEE Transactions on Pattern Analysis and Machine Intelligence. 35 pages, 2 figures, 33 tables
