208 Hits in 3.8 sec

Stochastic Segmentation Networks: Modelling Spatially Correlated Aleatoric Uncertainty [article]

Miguel Monteiro, Loïc Le Folgoc, Daniel Coelho de Castro, Nick Pawlowski, Bernardo Marques, Konstantinos Kamnitsas, Mark van der Wilk, Ben Glocker
2020 arXiv   pre-print
In this paper, we introduce stochastic segmentation networks (SSNs), an efficient probabilistic method for modelling aleatoric uncertainty with any image segmentation network architecture.  ...  SSNs outperform state-of-the-art for modelling correlated uncertainty in ambiguous images while being much simpler, more flexible, and more efficient.  ...  Discussion This paper introduces an efficient approach for modelling spatially correlated aleatoric uncertainty in segmentation.  ... 
arXiv:2006.06015v2 fatcat:pdh7d57hmve2bhrkesh27q3otu

Calibrated Adversarial Refinement for Stochastic Semantic Segmentation [article]

Elias Kassapis, Georgi Dikov, Deepak K. Gupta, Cedric Nugteren
2021 arXiv   pre-print
probability distribution over semantic classes and (ii) an adversarially trained stochastic network is used to model the inter-pixel correlations to refine the output of the first network into coherent  ...  To this end, we propose a novel two-stage, cascaded approach for calibrated adversarial refinement: (i) a standard segmentation network is trained with categorical cross entropy to predict a pixelwise  ...  We expect that increasing the number of samples from the refinement network will improve the aleatoric uncertainty estimates.  ... 
arXiv:2006.13144v3 fatcat:okytkfiusjcx7dmb2kn7abgyai

Towards Safe Deep Learning: Accurately Quantifying Biomarker Uncertainty in Neural Network Predictions [chapter]

Zach Eaton-Rosen, Felix Bragman, Sotirios Bisdas, Sébastien Ourselin, M. Jorge Cardoso
2018 Lecture Notes in Computer Science  
In this work we propose to use Bayesian neural networks to quantify uncertainty within the domain of semantic segmentation.  ...  When applied to a tumour volume estimation application, we demonstrate that by using such modelling of uncertainty, deep learning systems can be made to report volume estimates with well-calibrated error-bars  ...  In this work, we utilise a Bayesian deep learning model to measure the uncertainty of image segmentation. We fit network architectures with differing levels of stochasticity.  ... 
doi:10.1007/978-3-030-00928-1_78 fatcat:y44tznh7cffq7eqfql4bg3os4m

Aleatoric uncertainty estimation with test-time augmentation for medical image segmentation with convolutional neural networks

Guotai Wang, Wenqi Li, Michael Aertsen, Jan Deprest, Sébastien Ourselin, Tom Vercauteren
2019 Neurocomputing  
., model (epistemic) and image-based (aleatoric) uncertainties.  ...  We compare and combine our proposed aleatoric uncertainty with model uncertainty.  ...  In conclusion, we analyzed different types of uncertainties for CNN-based medical image segmentation by comparing and combining model (epistemic) and input-based (aleatoric) uncertainties.  ... 
doi:10.1016/j.neucom.2019.01.103 fatcat:hyw3t5peefhmjmbxdpxlwfyzjy

Towards Safe Autonomous Driving: Capture Uncertainty in the Deep Neural Network For Lidar 3D Vehicle Detection [article]

Di Feng, Lars Rosenbaum, Klaus Dietmayer
2018 arXiv   pre-print
Previous object detectors driven by deep learning do not explicitly model uncertainties in the neural network.  ...  The results also show that we can improve the detection performance by 1%-5% by modeling the aleatoric uncertainty.  ...  Fig. 8 : 8 An exemplary evolution of aleatoric and epistemic spatial uncertainty in sequential detections. PCC: Pearson correlation coefficient.  ... 
arXiv:1804.05132v2 fatcat:jwo2xiz4njd5ndkbh2mz5hvow4

The Hidden Uncertainty in a Neural Networks Activations [article]

Janis Postels, Hermann Blum, Yannick Strümpler, Cesar Cadena, Roland Siegwart, Luc Van Gool, Federico Tombari
2021 arXiv   pre-print
This work investigates whether this distribution moreover correlates with a model's epistemic uncertainty, thus indicates its ability to generalise to novel inputs.  ...  We verify our findings on both classification and regression models.  ...  Aleatoric uncertainty is assessed by its correlation with the prediction error on the training data distribution.  ... 
arXiv:2012.03082v2 fatcat:2txmq45dhbaytgoc7vn6dyvwuu

Bayesian Multi Scale Neural Network for Crowd Counting [article]

Abhinav Sagar
2022 arXiv   pre-print
On evaluating on ShanghaiTech, UCF-CC-50 and UCF-QNRF datasets using MSE and MAE as evaluation metrics, our network outperforms previous state of the art approaches while giving uncertainty estimates in  ...  Convolutional Neural Networks based on estimating the density map over the image has been highly successful in this domain.  ...  Uncertainty Estimation There are two main sources of uncertainty in model predictions: epistemic uncertainty is uncertainty due to our lack of knowledge and aleatoric uncertainty is due to stochasticity  ... 
arXiv:2007.14245v3 fatcat:5u2shi2r3veehowasscp4vvdny

Integrating Uncertainty in Deep Neural Networks for MRI based stroke analysis

Lisa Herzog, Elvis Murina, Oliver Dürr, Susanne Wegener, Beate Sick
2020 Medical Image Analysis  
Those methods take advantage of the uncertainty in the image predictions and report model uncertainty at the patient-level.  ...  Integrating uncertainty information about image predictions in aggregation models resulted in higher uncertainty measures to false patient classifications, which enabled to filter critical patient diagnoses  ...  Aleatoric uncertainty models the noise inherent in the input data itself.  ... 
doi:10.1016/ pmid:32801096 fatcat:35saaohqufhrhflmdcoocspcdm

Calibration and Uncertainty Quantification of Bayesian Convolutional Neural Networks for Geophysical Applications [article]

Lukas Mosser, Ehsan Zabihi Naeini
2021 arXiv   pre-print
We compare three different approaches to obtaining probabilistic models based on convolutional neural networks in a Bayesian formalism, namely Deep Ensembles, Concrete Dropout, and Stochastic Weight Averaging-Gaussian  ...  the popular Dropout technique to approximate Bayesian neural networks, and finally, we apply SWAG, a recent method that is based on the Bayesian inference equivalence of mini-batch Stochastic Gradient  ...  , yet we also observe high aleatoric uncertainties which can be attributed to irreducible model uncertainties.  ... 
arXiv:2105.12115v1 fatcat:5ah5twtcyjc7fbm2xil7q4ln3q

Neighborhood Spatial Aggregation MC Dropout for Efficient Uncertainty-aware Semantic Segmentation in Point Clouds [article]

Chao Qi, Jianqin Yin
2021 arXiv   pre-print
Uncertainty-aware semantic segmentation of the point clouds includes the predictive uncertainty estimation and the uncertainty-guided model optimization.  ...  The aleatoric uncertainty is integrated into the loss function to penalize noisy points, avoiding the over-fitting of the model to some degree.  ...  Our work achieves uncertainty-aware segmentation by working with existing PCSS networks.  ... 
arXiv:2201.07676v1 fatcat:g2tfg2w7lzex3opgfrsx3jou2q

Informative and Reliable Tract Segmentation for Preoperative Planning

Oeslle Lucena, Pedro Borges, Jorge Cardoso, Keyoumars Ashkan, Rachel Sparks, Sebastien Ourselin
2022 Frontiers in Radiology  
We demonstrate a high positive correlation between volume variance and segmentation errors, which indicates a good measure of reliability for tract segmentation ad uncertainty estimation.  ...  We use a 3D U-Net to segment white matter tracts. We then estimate model and data uncertainty using test time dropout and test time augmentation, respectively.  ...  Aleatoric Uncertainty Modeling We modeled aleatoric uncertainty using TTA.  ... 
doi:10.3389/fradi.2022.866974 fatcat:bqrh3oxsnrbhheja325hdhxpny

A Multi-Channel Uncertainty-Aware Multi-Resolution Network for MR to CT Synthesis

Kerstin Klaser, Pedro Borges, Richard Shaw, Marta Ranzini, Marc Modat, David Atkinson, Kris Thielemans, Brian Hutton, Vicky Goh, Gary Cook, Jorge Cardoso, Sebastien Ourselin
2021 Applied Sciences  
We propose the use of an uncertainty-aware multi-channel multi-resolution 3D cascade network specifically aiming for whole-body MR to CT synthesis.  ...  However, synthesising whole-body images remains largely uncharted territory, involving many challenges, including large image size and limited field of view, complex spatial context, and anatomical differences  ...  It is evident that in our setting the aleatoric uncertainty should be modelled as heteroscedastic, as task performance is expected to vary spatially due to the presence of artefacts, tissue boundaries,  ... 
doi:10.3390/app11041667 pmid:33763236 pmcid:PMC7610395 fatcat:j7znmvlparhcbnvrnxte4nnzua

Automatic Segmentation of Gross Target Volume of Nasopharynx Cancer using Ensemble of Multiscale Deep Neural Networks with Spatial Attention [article]

Haochen Mei, Wenhui Lei, Ran Gu, Shan Ye, Zhengwentai Sun, Shichuan Zhang, Guotai Wang
2021 arXiv   pre-print
We also estimate the uncertainty of segmentation results based on our model ensemble, which is of great importance for indicating the reliability of automatic segmentation results for radiotherapy planning  ...  Furthermore, we propose a spatial attention module to enable the network to focus on small target, and use channel attention to further improve the segmentation performance.  ...  [31] estimated lumen segmentation uncertainty for realistic patient-specific blood flow modeling. For deep CNNs, both epistemic and aleatoric uncertainty have been investigated in recent years.  ... 
arXiv:2101.11254v1 fatcat:mps5b7y7ubeari5wre7tbyahfy

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

Tinu Theckel Joy, Suman Sedai, Rahil Garnavi
2021 arXiv   pre-print
We investigate epistemic and aleatoric uncertainty capturing model confidence and data uncertainty respectively.  ...  We finally analyze the correlation between segmentation uncertainty and accuracy.  ...  We model the segmentation uncertainty using both epistemic and aleatoric uncertainty measures. We evaluate the generalization performance of the model for drusen segmentation.  ... 
arXiv:2101.08888v2 fatcat:rmys6xrrovek7nqvvyoj372wyu

SalsaNext: Fast, Uncertainty-aware Semantic Segmentation of LiDAR Point Clouds for Autonomous Driving [article]

Tiago Cortinhal, George Tzelepis, Eren Erdal Aksoy
2020 arXiv   pre-print
We finally inject a Bayesian treatment to compute the epistemic and aleatoric uncertainties for each point in the cloud.  ...  In this paper, we introduce SalsaNext for the uncertainty-aware semantic segmentation of a full 3D LiDAR point cloud in real-time.  ...  loss [2] . • To further estimate the epistemic (model) and aleatoric (observation) uncertainties for each 3D LiDAR point, the deterministic SalsaNet model was transformed into a stochastic format by  ... 
arXiv:2003.03653v3 fatcat:iwhvwniwizbpvonksrlnknc45e
« Previous Showing results 1 — 15 out of 208 results