Skip to content

U-Net + Diffusion2D UQ

Notebook: UNet_Diffusion2D_UQ_Tutorial.ipynb

This tutorial uses UNet2D for a 2D diffusion field-to-field mapping. The main UQ path is MC Dropout, while the tutorial also discusses how the same backbone can be paired with ensemble uncertainty for stronger multi-model diversity.

Primary references:

  • Ronneberger et al. (2015), U-Net: Convolutional Networks for Biomedical Image Segmentation
  • Lakshminarayanan et al. (2017), Simple and Scalable Predictive Uncertainty Estimation using Deep Ensembles