Bayesian diffusion models memorize training data when mutual information between restricted observations and training data exceeds log dataset size, and generalize otherwise.
Locality in image diffusion models emerges from data statistics
6 Pith papers cite this work. Polarity classification is still indexing.
6
Pith papers citing it
representative citing papers
Ambient Diffusion Policy enables better imitation learning from suboptimal robot data by leveraging spectral properties to restrict data usage to specific diffusion times.
Derives optimal score functions for diffusion models as wavelet expansions in terms of data moments, enabling architecture-agnostic analysis of which distribution attributes matter for denoising.
Diffusion models overfit denoising loss at intermediate noise but generalize in inference as model error smooths the flow field and sampling paths avoid memorized noisy training data.
citing papers explorer
-
An exact information theory of generalization phase transitions in Bayesian diffusion models
Bayesian diffusion models memorize training data when mutual information between restricted observations and training data exceeds log dataset size, and generalize otherwise.
-
Ambient Diffusion Policy: Imitation Learning from Suboptimal Data in Robotics
Ambient Diffusion Policy enables better imitation learning from suboptimal robot data by leveraging spectral properties to restrict data usage to specific diffusion times.
-
Where the Score Lives: A Wavelet View of Diffusion
Derives optimal score functions for diffusion models as wavelet expansions in terms of data moments, enabling architecture-agnostic analysis of which distribution attributes matter for denoising.
-
Diffusion Models Memorize in Training -- and Generalize in Inference
Diffusion models overfit denoising loss at intermediate noise but generalize in inference as model error smooths the flow field and sampling paths avoid memorized noisy training data.
- Streamlining Analysis and Design of Two-Dimensional Electronic Spectroscopy using Machine Learning
- Score-based Membership Inference on Diffusion Models