Classifying by Proxy: Explainable and Reproducible Ensemble of Proxy Tasks for Child Sexual Abuse Imagery Classification

2026-06-14

AI Summary

This study addresses the challenges of reproducibility and interpretability in child sexual abuse imagery (CSAI) classification, which are hindered by data sensitivity and opaque model decisions. To overcome these limitations, the work introduces, for the first time, an integrated proxy-task framework for real-world CSAI classification that preserves data privacy while enhancing both performance and interpretability. By carefully selecting multiple relevant proxy tasks and refining the training architecture, the proposed method achieves a balanced accuracy of 91.9% on the RCPD dataset—outperforming state-of-the-art representation learning models such as DINO. Moreover, it provides interpretable justifications for its classification decisions, thereby substantially improving model safety, reproducibility, and transparency.

Abstract

Child Sexual Abuse Imagery (CSAI) classification systems are needed solutions for lessening the psychological impacts often felt by law enforcement agents responsible for evaluating these materials and for efficient removal of these materials from the web. However, due to the nature of the task, researching and developing such systems is not a trivial endeavor. The images are highly sensitive, and the related datasets are under restrictive access regimes, which means most studies in the area are not reproducible or distributable and are therefore hard to compare and validate. More concerning still, most models for this task today lack an aspect often desired by law enforcement agents: explainability. In this paper, we apply an ensemble of Proxy Tasks -- tasks that correlate to CSAI classification -- yielding improvements in reproducibility, explainability, and security for distribution. This concept is applied for the first time to real CSAI, with a novel selection of relevant Proxy Tasks (selected from the CSAI literature) and training adaptations to the original framework. Our final model achieves competitive results, yielding 91.9% balanced accuracy on the RCPD dataset with the best Proxy Task combination. We furthermore contrast these results with the best-in-class representation learning model, DINO, and show that our ensemble improves accuracy and provides explanations for its classification results, a feature that a single deep learning model can seldom provide.

Problem

• Child Sexual Abuse Imagery
• reproducibility
• explainability
• sensitive data
• classification

Innovation

• Proxy Tasks
• Explainable AI
• Ensemble Learning
• CSAI Classification
• Reproducibility