Accelerating Anchors via Specialization and Feature Transformation

TL;DR

This work tackles the computational bottleneck of Anchors by proposing an offline online hybrid strategy: offline pre-training builds general, high-coverage anchors from $N$ representative inputs, while online refinement uses horizontal and vertical rule transformations to adapt and specific-ify these anchors for new inputs. The horizontal transformation maps pre-trained predicates to similar online features, and the vertical transformation incrementally strengthens the rule until the desired precision is met, leveraging Anchors' iterative sampling. Empirical results across tabular, text, and image domains show substantial time savings (up to over 270% acceleration) while preserving precision comparable to Anchors, demonstrating practical applicability for real-time explanations. The approach enables scalable, model-agnostic explanations that retain interpretability and fidelity, broadening Anchors' usability in sensitive, real-time decision contexts.

Abstract

Anchors is a popular local model-agnostic explanation technique whose applicability is limited by its computational inefficiency. To address this limitation, we propose a pre-training-based approach to accelerate Anchors without compromising the explanation quality. Our approach leverages the iterative nature of Anchors' algorithm which gradually refines an explanation until it is precise enough for a given input by providing a general explanation that is obtained through pre-training as Anchors' initial explanation. Specifically, we develop a two-step rule transformation process: the horizontal transformation adapts a pre-trained explanation to the current input by replacing features, and the vertical transformation refines the general explanation until it is precise enough for the input. We evaluate our method across tabular, text, and image datasets, demonstrating that it significantly reduces explanation generation time while maintaining fidelity and interpretability, thereby enabling the practical adoption of Anchors in time-sensitive applications.

Figures (4)

• Figure 1: An example of our workflow for image.
• Figure 2: The overall workflow of Anchors.
• Figure 3: The absolute time cost comparison before and after acceleration.(Sentiment Analysis)
• Figure 4: The fidelity of our method and Anchors.(Income Prediction)