Minimizing Risk Through Minimizing Model-Data Interaction: A Protocol For Relying on Proxy Tasks When Designing Child Sexual Abuse Imagery Detection Models

TL;DR

This paper introduces a protocol for designing CSAI detection models using proxy tasks to minimize direct data interaction, enhancing privacy and security.

Contribution

It formalizes the concept of proxy tasks for CSAI detection and demonstrates their effective use in a real-world application without training on sensitive data.

Findings

Achieved promising results on a real-world CSAI dataset.

First application of proxy tasks in CSAI detection.

Demonstrated that models can perform well without direct access to sensitive data.

Abstract

The distribution of child sexual abuse imagery (CSAI) is an ever-growing concern of our modern world; children who suffered from this heinous crime are revictimized, and the growing amount of illegal imagery distributed overwhelms law enforcement agents (LEAs) with the manual labor of categorization. To ease this burden researchers have explored methods for automating data triage and detection of CSAI, but the sensitive nature of the data imposes restricted access and minimal interaction between real data and learning algorithms, avoiding leaks at all costs. In observing how these restrictions have shaped the literature we formalize a definition of "Proxy Tasks", i.e., the substitute tasks used for training models for CSAI without making use of CSA data. Under this new terminology we review current literature and present a protocol for making conscious use of Proxy Tasks together with consistent input from LEAs to design better automation in this field. Finally, we apply this protocol to study -- for the first time -- the task of Few-shot Indoor Scene Classification on CSAI, showing a final model that achieves promising results on a real-world CSAI dataset whilst having no weights actually trained on sensitive data.