LLMLow
Overfit Kaggle Models Discouraged Data Science Competitors
- Occurred
- May 1, 2017
- Published
- May 1, 2017
- Source
- AI Incident Database
Summary
In the “The Nature Conservancy Fisheries Monitoring” competition on the data science competition website Kaggle, a number of competitors overfit their image classifier models to a poorly representative validation data set.
Developers
- Individual Kaggle Competitors
Deployers
- Individual Kaggle Competitors
Harmed parties
- Individual Kaggle Competitors
Classification
Category confidence: 70%
Severity confidence: 90%
Rationale: Overfitting in Kaggle competition is a technical failure, not a safety incident, fitting LOW severity.
Classifier: Qwen/Qwen3.5-35B-A3B-FP8