BEE: Metric-Adapted Explanations via Baseline Exploration-Exploitation

TL;DR

This work tackles the lack of a universal evaluation metric and baseline representation in explainable AI by introducing Baseline Exploration-Exploitation (BEE). BEE treats the baseline as a learned random tensor drawn from a mixture of distributions and optimizes it through contextual exploration-exploitation to tailor explanations to a given metric. Explanations are produced via a path-integrated scheme over intermediate representations, yielding multiple maps across layers and selecting the best performing map per metric; the method supports pretrained and inference-time finetuning. Across ImageNet experiments with CNN and ViT backbones, BEE achieves state-of-the-art performance on a broad set of objective metrics, demonstrating strong generalization and the ability to adapt explanations to diverse evaluation criteria.

Abstract

Two prominent challenges in explainability research involve 1) the nuanced evaluation of explanations and 2) the modeling of missing information through baseline representations. The existing literature introduces diverse evaluation metrics, each scrutinizing the quality of explanations through distinct lenses. Additionally, various baseline representations have been proposed, each modeling the notion of missingness differently. Yet, a consensus on the ultimate evaluation metric and baseline representation remains elusive. This work acknowledges the diversity in explanation metrics and baselines, demonstrating that different metrics exhibit preferences for distinct explanation maps resulting from the utilization of different baseline representations and distributions. To address the diversity in metrics and accommodate the variety of baseline representations in a unified manner, we propose Baseline Exploration-Exploitation (BEE) - a path-integration method that introduces randomness to the integration process by modeling the baseline as a learned random tensor. This tensor follows a learned mixture of baseline distributions optimized through a contextual exploration-exploitation procedure to enhance performance on the specific metric of interest. By resampling the baseline from the learned distribution, BEE generates a comprehensive set of explanation maps, facilitating the selection of the best-performing explanation map in this broad set for the given metric. Extensive evaluations across various model architectures showcase the superior performance of BEE in comparison to state-of-the-art explanation methods on a variety of objective evaluation metrics.

Figures (6)

• Figure 1: ViT qualitative results: Explanation maps produced using ViT-B w.r.t. the classes (top to bottom): ‘macaque’, ‘porcupine, hedgehog’, ‘alp’ and ‘planetarium’.
• Figure 2: Metric score vs. number of drawn baselines. A comprehensive comparison among fBEE, pBEE, nBEE, and various types of baseline distributions is presented for each metric using the RN model.
• Figure 3: A comprehensive comparison among fBEE, pBEE, nBEE, and various types of baseline distributions is presented for each metric using the RN model. x axis - number of drawn baselines, y axis - metric score. See Section \ref{['subsec:results']} for details.
• Figure 4: Explanation maps produced for the normal and blur baseline types using RN w.r.t. 'baboon' and 'basenji'. On the finetune phase over the PIC metric, BEE favored the normal baseline for the 'baboon' and the blur baseline for the 'basenji'. Both choices yielded better performance.
• Figure 5: The reward distribution varies across different metrics and distinct types of baseline distributions. For example, PIC, INS, and NEG favor the Normal baseline, while DEL, POS, and SIC favor the Uniform baseline. The observed preferences of different metrics for distinct baselines underscore the necessity of the adaptive sampling employed by BEE.