A theoretical basis for model collapse in recursive training
arXiv:2506.09401v41 citationsh-index: 3
Originality Incremental advance
AI Analysis
This provides theoretical insights into a fundamental issue in machine learning, particularly for generative AI, but is incremental as it builds on known collapse phenomena.
The paper investigates the phenomenon of model collapse in recursive training of generative models, showing that the asymptotic behavior diverges based on the presence of an external sample source, even if minor.
It is known that recursive training from generative models can lead to the so called `collapse' of the simulated probability distribution. This note shows that one in fact gets two different asymptotic behaviours depending on whether an external source, howsoever minor, is also contributing samples.