The Directed Prediction Change - Efficient and Trustworthy Fidelity Assessment for Local Feature Attribution Methods

TL;DR

This paper introduces the Directed Prediction Change (DPC), a deterministic and efficient metric for evaluating the fidelity of local feature attribution methods, crucial for trustworthy explanations in high-stakes settings.

Contribution

The paper proposes DPC, a novel, faster, and deterministic fidelity metric that improves upon existing methods by incorporating directionality in perturbations and attributions.

Findings

DPC achieves nearly tenfold speedup over traditional metrics.

DPC provides deterministic and reproducible fidelity assessments.

Evaluation on diverse datasets and models confirms DPC's effectiveness.

Abstract

The utility of an explanation method critically depends on its fidelity to the underlying machine learning model. Especially in high-stakes medical settings, clinicians and regulators require explanations that faithfully reflect the model's decision process. Existing fidelity metrics such as Infidelity rely on Monte Carlo approximation, which demands numerous model evaluations and introduces uncertainty due to random sampling. This work proposes a novel metric for evaluating the fidelity of local feature attribution methods by modifying the existing Prediction Change (PC) metric within the Guided Perturbation Experiment. By incorporating the direction of both perturbation and attribution, the proposed Directed Prediction Change (DPC) metric achieves an almost tenfold speedup and eliminates randomness, resulting in a deterministic and trustworthy evaluation procedure that measures the same property as local Infidelity. DPC is evaluated on two datasets (skin lesion images and financial tabular data), two black-box models, seven explanation algorithms, and a wide range of hyperparameters. Across $4\,744$ distinct explanations, the results demonstrate that DPC, together with PC, enables a holistic and computationally efficient evaluation of both baseline-oriented and local feature attribution methods, while providing deterministic and reproducible outcomes.