Deep Model Interpretation with Limited Data : A Coreset-based Approach

TL;DR

This paper introduces a coreset-based framework for model interpretation that reduces computational costs by selecting representative data subsets, enabling more efficient insights into trained models without sacrificing interpretability.

Contribution

It proposes a novel coreset selection approach combined with a similarity-based evaluation protocol to improve and assess model interpretation methods on large datasets.

Findings

Coreset-based sampling maintains interpretation quality with fewer data points.

The framework improves computational efficiency of interpretation methods.

Experimental results validate the effectiveness across multiple models and datasets.

Abstract

Model Interpretation aims at the extraction of insights from the internals of a trained model. A common approach to address this task is the characterization of relevant features internally encoded in the model that are critical for its proper operation. Despite recent progress of these methods, they come with the weakness of being computationally expensive due to the dense evaluation of datasets that they require. As a consequence, research on the design of these methods have focused on smaller data subsets which may led to reduced insights. To address these computational costs, we propose a coreset-based interpretation framework that utilizes coreset selection methods to sample a representative subset of the large dataset for the interpretation task. Towards this goal, we propose a similarity-based evaluation protocol to assess the robustness of model interpretation methods towards the amount data they take as input. Experiments considering several interpretation methods, DNN models, and coreset selection methods show the effectiveness of the proposed framework.