Our paper "When Sample Selection Bias Precipitates Model Collapse" is accepted by ICML 2026

We are glad to share that our paper "When Sample Selection Bias Precipitates Model Collapse" has been accepted to the International Conference on Machine Learning (ICML 2026).

The paper analyzes how sample selection bias can precipitate model collapse — the progressive quality loss that can occur when models are trained on selectively filtered or model-generated data — identifying the conditions under which collapse emerges and how it can be mitigated.