449 Hits in 4.4 sec

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
We analyse epistemic and aleatoric uncertainty inferred from the representations of different layers and conclude that deeper layers lead to uncertainty with similar behaviour as established - but computationally  ...  The distribution of a neural network's latent representations has been successfully used to detect out-of-distribution (OOD) data.  ...  In conclusion, we find that the hidden activations of neural networks contain information about both aleatoric and epistemic uncertainty.  ... 
arXiv:2012.03082v2 fatcat:2txmq45dhbaytgoc7vn6dyvwuu

Deep Deterministic Uncertainty: A Simple Baseline [article]

Jishnu Mukhoti, Andreas Kirsch, Joost van Amersfoort, Philip H.S. Torr, Yarin Gal
2022 arXiv   pre-print
This conceptually simple *Deep Deterministic Uncertainty (DDU)* baseline can also be used to disentangle aleatoric and epistemic uncertainty and performs as well as Deep Ensembles, the state-of-the art  ...  * DUQ and SNGP's epistemic uncertainty predictions using simple Gaussian Discriminant Analysis *post-training* as a separate feature-space density estimator -- without fine-tuning on OoD data, feature  ...  Deep Deterministic Uncertainty As introduced in §1, we propose to use a deterministic neural network with an appropriately regularized featurespace, using spectral normalization (Liu et al., 2020a) ,  ... 
arXiv:2102.11582v3 fatcat:tt75wgvfxvdnbh3vabtj62rjwy


M. Mehltretter
2022 ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences  
Moreover, the evaluation reveals the importance of considering both, aleatoric and epistemic uncertainty, in order to achieve an accurate estimation of the overall uncertainty related to a depth estimate  ...  To approach this objective, a holistic method to jointly estimate disparity and uncertainty is presented in this work, taking into account both aleatoric and epistemic uncertainty.  ...  Common realisations of stochastic neural networks are ensembles of deterministic neural networks that have been trained independently, Monte Carlo dropout and BNNs.  ... 
doi:10.5194/isprs-annals-v-2-2022-69-2022 fatcat:v4glmtyqe5at3ne7yjenbxov6a

Estimating Risk and Uncertainty in Deep Reinforcement Learning [article]

William R. Clements, Bastien Van Delft, Benoît-Marie Robaglia, Reda Bahi Slaoui, Sébastien Toth
2020 arXiv   pre-print
Epistemic uncertainty stems from limited data and is useful for exploration, whereas aleatoric uncertainty arises from stochastic environments and must be accounted for in risk-sensitive applications.  ...  We highlight the challenges involved in simultaneously estimating both of them, and propose a framework for disentangling and estimating these uncertainties on learned Q-values.  ...  However, in the limit of infinite width Bayesian neural networks are uncorrelated for normal priors and separable likelihoods (Neal, 1995) .  ... 
arXiv:1905.09638v5 fatcat:iq4savzrgfdqlovzzkowgfkdbe


M. Mehltretter
2020 ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences  
Based on the well-known and commonly employed GC-Net architecture, a novel probabilistic neural network is presented, for the task of joint depth and uncertainty estimation from epipolar rectified stereo  ...  The quality of the estimated depth and uncertainty information is assessed in an extensive evaluation on three different datasets.  ...  and TU Braunschweig and by the NVIDIA Corporation with the donation of the Titan V GPU used for this research.  ... 
doi:10.5194/isprs-annals-v-2-2020-161-2020 fatcat:ks32kkzo3zgzfprfne27xhmdze

BaCOUn: Bayesian Classifers with Out-of-Distribution Uncertainty [article]

Théo Guénais, Dimitris Vamvourellis, Yaniv Yacoby, Finale Doshi-Velez, Weiwei Pan
2020 arXiv   pre-print
We propose a Bayesian framework to obtain reliable uncertainty estimates for deep classifiers.  ...  Our approach consists of a plug-in "generator" used to augment the data with an additional class of points that lie on the boundary of the training data, followed by Bayesian inference on top of features  ...  We implement deterministic Neural Networks usingPy- Torch (Paszke et al., 2019).  ... 
arXiv:2007.06096v1 fatcat:ejzuirw7gveyhfaemp4toqkcni

Uncertainty Estimation for End-To-End Learned Dense Stereo Matching via Probabilistic Deep Learning [article]

Max Mehltretter
2020 arXiv   pre-print
Based on the well-known and commonly employed GC-Net architecture, a novel probabilistic neural network is presented, for the task of joint depth and uncertainty estimation from epipolar rectified stereo  ...  The quality of the estimated depth and uncertainty information is assessed in an extensive evaluation on three different datasets.  ...  and TU Braunschweig and by the NVIDIA Corporation with the donation of the Titan V GPU used for this research.  ... 
arXiv:2002.03663v1 fatcat:ne44thf53fguzhot7lznnrzc4m

Hybrid Process Models in Electrochemical Syntheses under Deep Uncertainty

Fenila Francis-Xavier, Fabian Kubannek, René Schenkendorf
2021 Processes  
Still, standard parameter sensitivity concepts may fail to address comprehensive parameter uncertainty problems, i.e., deep uncertainty with aleatoric and epistemic contributions.  ...  The general findings of the hybrid model sensitivity studies under deep uncertainty are twofold. First, epistemic uncertainty has a significant effect on uncertainty analysis.  ...  and the consideration of aleatoric and epistemic uncertainties in the context of deep uncertainty.  ... 
doi:10.3390/pr9040704 fatcat:4bmruls3djb73n2cgefuccpsha

Success of Uncertainty-Aware Deep Models Depends on Data Manifold Geometry [article]

Mark Penrod, Harrison Termotto, Varshini Reddy, Jiayu Yao, Finale Doshi-Velez, Weiwei Pan
2022 arXiv   pre-print
Our finding suggests an interesting direction in the study of uncertainty-aware deep learning models.  ...  Thus, we compare six uncertainty-aware deep learning models on a set of edge-case tasks: robustness to adversarial attacks as well as out-of-distribution and adversarial detection.  ...  Uncertainty estimation using a single deep deterministic neural network, 2020. Wenzel, F., Snoek, J., Tran, D., and Jenatton, R.  ... 
arXiv:2208.01705v2 fatcat:ynnqr3q7wbawlfphsqdg34yqiq

A Review and Comparative Study on Probabilistic Object Detection in Autonomous Driving [article]

Di Feng, Ali Harakeh, Steven Waslander, Klaus Dietmayer
2020 arXiv   pre-print
However, there is no summary on uncertainty estimation in deep object detection, and existing methods are not only built with different network architectures and uncertainty estimation methods, but also  ...  First, we provide an overview of generic uncertainty estimation in deep learning, and then systematically survey existing methods and evaluation metrics for probabilistic object detection.  ...  Epistemic and Aleatoric uncertainty Predictive uncertainty in deep neural networks can be decomposed into epistemic uncertainty and aleatoric uncertainty [57] .  ... 
arXiv:2011.10671v1 fatcat:a7exswrvjfczpln7xojouzklby

Deep Reinforcement Learning in a Handful of Trials using Probabilistic Dynamics Models [article]

Kurtland Chua and Roberto Calandra and Rowan McAllister and Sergey Levine
2018 arXiv   pre-print
This is especially true with high-capacity parametric function approximators, such as deep networks. In this paper, we study how to bridge this gap, by employing uncertainty-aware dynamics models.  ...  We propose a new algorithm called probabilistic ensembles with trajectory sampling (PETS) that combines uncertainty-aware deep network dynamics models with sampling-based uncertainty propagation.  ...  An advantage of using TS∞ is that aleatoric and epistemic uncertainties are separable [Depeweg et al., 2018] .  ... 
arXiv:1805.12114v2 fatcat:bqyf43uyrbao5mzaojrwgtrb5u

Aleatoric and Epistemic Uncertainty with Random Forests [chapter]

Mohammad Hossein Shaker, Eyke Hüllermeier
2020 Lecture Notes in Computer Science  
In this regard, we also compare random forests with deep neural networks, which have been used for a similar purpose.  ...  In particular, the idea of distinguishing between two important types of uncertainty, often refereed to as aleatoric and epistemic, has recently been studied in the setting of supervised learning.  ...  In particular, a distinction between aleatoric and epistemic uncertainty has been advocated in the literature on deep learning [6] , where the limited awareness of neural networks of their own competence  ... 
doi:10.1007/978-3-030-44584-3_35 fatcat:e4pgfdjj4rhejgrqzacf277jbm

AutoDEUQ: Automated Deep Ensemble with Uncertainty Quantification [article]

Romain Egele, Romit Maulik, Krishnan Raghavan, Bethany Lusch, Isabelle Guyon, Prasanna Balaprakash
2022 arXiv   pre-print
We use the law of total variance to decompose the predictive variance of deep ensembles into aleatoric (data) and epistemic (model) uncertainties.  ...  Deep neural networks are powerful predictors for a variety of tasks. However, they do not capture uncertainty directly.  ...  theoretical insights into the quality of epistemic uncertainty under the various data generation assumptions. V. ACKNOWLEDGEMENT  ... 
arXiv:2110.13511v3 fatcat:wfol5b7pdba7xjf5q4ojsecc6q

Bayesian deep learning of affordances from RGB images [article]

Lorenzo Mur-Labadia, Ruben Martinez-Cantin
2021 arXiv   pre-print
Our Bayesian model is able to capture both the aleatoric uncertainty from the scene and the epistemic uncertainty associated with the model and previous learning process.  ...  For comparison, we estimate the uncertainty using two state-of-the-art techniques: Monte Carlo dropout and deep ensembles. We also compare different types of CNN encoders for feature extraction.  ...  The posterior distribution obtained in both method can be further separated in aleatoric and epistemic uncertainty.  ... 
arXiv:2109.12845v1 fatcat:n7gqlxomina75msjhret3gak7i

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 proposed probabilistic detector represents reliable epistemic uncertainty and aleatoric uncertainty in classification and localization tasks.  ...  ACKNOWLEDGMENT We thank Zhongyu Lou and Florian Faion for their suggestions and inspiring discussions.  ... 
arXiv:1804.05132v2 fatcat:jwo2xiz4njd5ndkbh2mz5hvow4
« Previous Showing results 1 — 15 out of 449 results