论文标题
效率与一致性之间的权衡，以取消删除解释
Trade-off Between Efficiency and Consistency for Removal-based Explanations
论文作者
论文摘要
在当前的解释方法的景观中，大多数主要方法（例如Shap and Lime）采用基于删除的技术来评估单个特征的影响，通过模拟各种方案中使用省略的特定特征。尽管如此，这些方法主要强调原始环境中的效率，通常会导致一般不一致。在本文中，我们证明了这种不一致是这些方法的固有方面，通过建立不可能的三位一体定理，这表明可以同时保持这种可解释性，效率和一致性。认识到达到理想解释仍然难以捉摸，我们提出将解释误差作为衡量效率和不一致的指标。为此，我们提出了基于标准多项式基础建立的两种新型算法，旨在最大程度地减少解释误差。我们的经验发现表明，与替代技术相比，所提出的方法的解释误差大大减少了31.8倍。代码可在https://github.com/trusty-ai/felfcited-consistent-explanation上找到。
In the current landscape of explanation methodologies, most predominant approaches, such as SHAP and LIME, employ removal-based techniques to evaluate the impact of individual features by simulating various scenarios with specific features omitted. Nonetheless, these methods primarily emphasize efficiency in the original context, often resulting in general inconsistencies. In this paper, we demonstrate that such inconsistency is an inherent aspect of these approaches by establishing the Impossible Trinity Theorem, which posits that interpretability, efficiency, and consistency cannot hold simultaneously. Recognizing that the attainment of an ideal explanation remains elusive, we propose the utilization of interpretation error as a metric to gauge inefficiencies and inconsistencies. To this end, we present two novel algorithms founded on the standard polynomial basis, aimed at minimizing interpretation error. Our empirical findings indicate that the proposed methods achieve a substantial reduction in interpretation error, up to 31.8 times lower when compared to alternative techniques. Code is available at https://github.com/trusty-ai/efficient-consistent-explanations.
