Why Attentions May Not Be Interpretable?
원문 초록
Attention-based methods have played important roles in model interpretations, where the calculated attention weights are expected to highlight the critical parts of inputs~(e.g., keywords in sentences). However, recent research found that attention-as-importance interpretations often do not work as we expected. For example, learned attention weights sometimes highlight less meaningful tokens like "[SEP]", ",", and ".", and are frequently uncorrelated with other feature importance indicators like gradient-based measures. A recent debate over whether attention is an explanation or not has drawn considerable interest. In this paper, we demonstrate that one root cause of this phenomenon is the combinatorial shortcuts, which means that, in addition to the highlighted parts, the attention weights themselves may carry extra information that could be utilized by downstream models after attention layers. As a result, the attention weights are no longer pure importance indicators. We theoretically analyze combinatorial shortcuts, design one intuitive experiment to show their existence, and propose two methods to mitigate this issue. We conduct empirical studies on attention-based interpretation models. The results show that the proposed methods can effectively improve the interpretability of attention mechanisms.
함께 살펴볼 논문
공통 태그를 근거로 연결했습니다. 실제 인용 관계나 논문의 우수성을 뜻하지 않습니다.
Enhancing interpretability of rule-based classifiers through feature graphs
Profiling Electric Vehicles via Early Charging Voltage Patterns
The Reader is the Metric: How Textual Features and Reader Profiles Explain Conflicting Evaluations of AI Creative Writing
Multi-criteria Rank-based Aggregation for Explainable AI
EPIC: Explanation of Pretrained Image Classification Networks via Prototype
FairSHAP: Preprocessing for Fairness Through Attribution-Based Data Augmentation
토론
이 논문과 연결된 포럼 글이 아직 없습니다.