通过主动解释引导克服图神经网络中的捷径学习
Overcoming Shortcut Learning in Graph Neural Networks through Active Explanation Guidance
浏览论文内容
中文总结 AI 辅助
该研究针对GNN易利用非因果捷径导致分布外任务可靠性低的问题，提出架构无关的人在回路策略XIGL，结合主动学习优先选择含捷径的解释以降低反馈成本，经多种GNN架构验证其有效性。
中文摘要 AI 辅助
图神经网络（GNN）在解决预测任务时可能会无意中利用捷径，即那些与预测相关但并非预测因果因素的边、节点和特征，这会损害其在分布外任务中的可靠性。我们提出XIGL，一种架构无关的人在回路策略，用于去除GNN中的此类捷径。我们的核心见解有两点：一是可通过检查GNN的解释来检测对捷径的依赖；二是一旦意识到此类捷径，足够专业的用户可提供定制的纠正反馈，这有助于解耦模型。XIGL支持任何查询策略，但由于纠正反馈的获取成本较高，我们开发了一种主动学习策略，优先选择更可能显示捷径行为的解释，以降低标注和认知成本。我们在多种GNN架构上展示了XIGL的有效性，包括现有及提出的基于解释的策略，其实现可在线获取。
英文摘要
Graph Neural Networks (GNNs) can solve prediction tasks by unintentionally exploiting shortcuts---that is, edges, nodes, and features that correlate with but are not causal for the prediction---which compromise their reliability in out-of-distribution tasks. We introduce XIGL, an architecture-agnostic human-in-the-loop strategy for removing such shortcuts from GNNs. Our key insight is twofold. On the one hand, reliance on shortcuts can be detected by inspecting GNN explanations. On the other hand, once made aware of such shortcuts, sufficiently expert users can provide tailored corrective feedback, which helps deconfound the model. XIGL supports any query strategy; however, since corrective feedback can be expensive to acquire, we develop an active learning strategy for prioritizing explanations that are more likely to display shortcut behavior, lowering annotation and cognitive costs. We showcase the effectiveness of XIGL, including both existing and proposed explanation-based strategies, on several GNN architectures. Our implementation is available online.
发表机构
- VU Amsterdam（阿姆斯特丹自由大学）
- University of Trento（特伦托大学）
- UiT The Arctic University of Norway（挪威北极大学）
机构由 AI 辅助整理，请以论文原文为准。
