PCEvE: Part Contribution Evaluation Based Model Explanation for Human Figure Drawing Assessment and Beyond

TL;DR

The paper introduces PCEvE, a novel explanation framework for human figure drawing assessment that uses part contribution evaluation via Shapley Values to provide clear, comprehensive, and scalable model explanations beyond traditional pixel attribution methods.

Contribution

It proposes a new part contribution evaluation method that simplifies model explanations by using part-based Shapley Values, extending interpretability to class and task levels.

Findings

PCEvE effectively explains model decisions with part contribution histograms.

The method is validated on multiple datasets, including HFD and Stanford Cars.

PCEvE offers a scalable, interpretable alternative to pixel attribution approaches.

Abstract

For automatic human figure drawing (HFD) assessment tasks, such as diagnosing autism spectrum disorder (ASD) using HFD images, the clarity and explainability of a model decision are crucial. Existing pixel-level attribution-based explainable AI (XAI) approaches demand considerable effort from users to interpret the semantic information of a region in an image, which can be often time-consuming and impractical. To overcome this challenge, we propose a part contribution evaluation based model explanation (PCEvE) framework. On top of the part detection, we measure the Shapley Value of each individual part to evaluate the contribution to a model decision. Unlike existing attribution-based XAI approaches, the PCEvE provides a straightforward explanation of a model decision, i.e., a part contribution histogram. Furthermore, the PCEvE expands the scope of explanations beyond the conventional sample-level to include class-level and task-level insights, offering a richer, more comprehensive understanding of model behavior. We rigorously validate the PCEvE via extensive experiments on multiple HFD assessment datasets. Also, we sanity-check the proposed method with a set of controlled experiments. Additionally, we demonstrate the versatility and applicability of our method to other domains by applying it to a photo-realistic dataset, the Stanford Cars.