Nürnberg NLP system wins ChildSafeAds tasks with diverse voter ensembles
Nürnberg NLP at ChildSafeAds 2026: Structurally Dissimilar Voter Ensembles under Four Levels of Data Access
Computation and LanguageMachine Learning
Summary
Monitoring ads in videos for kids on YouTube is hard, especially when available data is limited. The Nürnberg NLP team built a system combining multiple different classifiers, each looking at the problem in its own way. By carefully selecting and combining these classifiers, the system performed best in two of the three categories tested, showing it can effectively spot ads and compliance issues in kids' videos. They also studied how having more or less data affects performance and the related costs.
What this means in practice
- •For content moderation teams: Improve automated detection of commercial content and compliance flags in child-directed videos using diverse ensembles tailored to data availability.
- •For advertising compliance teams: Deploy scalable monitoring tools that maintain performance across varying data access scenarios to ensure advertisements meet child safety standards.
Authors
Philipp Steigerwald, Eric Rudolph, Jens Albrecht
Abstract
We describe the Nürnberg NLP system for ChildSafeAds 2026. The shared task asks what a monitoring system for commercial content in child-facing YouTube videos can achieve at a given level of data access. We answer with per-subtask ensembles of nine voters, organised into three branches that differ in backbone, adaptation method and class scope. Selection rests on channel-disjoint cross-validation, with the development set as a transfer check. The system wins two of the three subtasks. Its product-category score (ST2, 0.8243) and its compliance-flag score (ST3, 0.6530) are the best of the 22 final entries, and it places third on the task mean (0.7079). We further compare four access levels and report the cost at test-set scale.