Modeling Multimodal Aleatoric Uncertainty in Segmentation with Mixture of Stochastic Expert

12/14/2022
by   Zhitong Gao, et al.
0

Equipping predicted segmentation with calibrated uncertainty is essential for safety-critical applications. In this work, we focus on capturing the data-inherent uncertainty (aka aleatoric uncertainty) in segmentation, typically when ambiguities exist in input images. Due to the high-dimensional output space and potential multiple modes in segmenting ambiguous images, it remains challenging to predict well-calibrated uncertainty for segmentation. To tackle this problem, we propose a novel mixture of stochastic experts (MoSE) model, where each expert network estimates a distinct mode of the aleatoric uncertainty and a gating network predicts the probabilities of an input image being segmented in those modes. This yields an efficient two-level uncertainty representation. To learn the model, we develop a Wasserstein-like loss that directly minimizes the distribution distance between the MoSE and ground truth annotations. The loss can easily integrate traditional segmentation quality measures and be efficiently optimized via constraint relaxation. We validate our method on the LIDC-IDRI dataset and a modified multimodal Cityscapes dataset. Results demonstrate that our method achieves the state-of-the-art or competitive performance on all metrics.

READ FULL TEXT

page 8

page 9

page 17

page 19

page 20

page 23

research
03/15/2023

Stochastic Segmentation with Conditional Categorical Diffusion Models

Semantic segmentation has made significant progress in recent years than...
research
12/09/2022

Reliable Multimodal Trajectory Prediction via Error Aligned Uncertainty Optimization

Reliable uncertainty quantification in deep neural networks is very cruc...
research
08/24/2020

Probabilistic Deep Learning for Instance Segmentation

Probabilistic convolutional neural networks, which predict distributions...
research
06/29/2022

Uncertainty-aware Panoptic Segmentation

Reliable scene understanding is indispensable for modern autonomous syst...
research
06/10/2020

Stochastic Segmentation Networks: Modelling Spatially Correlated Aleatoric Uncertainty

In image segmentation, there is often more than one plausible solution f...
research
01/21/2021

Analyzing Epistemic and Aleatoric Uncertainty for Drusen Segmentation in Optical Coherence Tomography Images

Age-related macular degeneration (AMD) is one of the leading causes of p...
research
06/09/2019

Overcoming Limitations of Mixture Density Networks: A Sampling and Fitting Framework for Multimodal Future Prediction

Future prediction is a fundamental principle of intelligence that helps ...

Please sign up or login with your details

Forgot password? Click here to reset