What Uncertainties Do We Need in Bayesian Deep Learning for Computer Vision?
Explore this paper's citation graph
Summary
A Bayesian deep learning framework combining input-dependent aleatoric uncertainty together with epistemic uncertainty is presented, which makes the loss more robust to noisy data, also giving new state-of-the-art results on segmentation and depth regression benchmarks.
- Type
- preprint
- Published
- 2017-03-15
- Cited by
- 6,536
- References
- 41
- Access
- Open access
- OpenAlex
- https://openalex.org/W2600383743
- Semantic Scholar
- https://api.semanticscholar.org/CorpusID:71134
Keywords
Uncertainty quantification, Artificial intelligence, Deep learning, Computer science, Bayesian probability
References
- Bayesian Convolutional Neural Networks with Bernoulli Approximate Variational Inference
- An Introduction to Variational Methods for Graphical Models
- Estimating the mean and variance of the target probability distribution
- Fully convolutional networks for semantic segmentation
- Predicting Depth, Surface Normals and Semantic Labels with a Common Multi-scale Convolutional Architecture
- SegNet: A Deep Convolutional Encoder-Decoder Architecture for Image Segmentation
- A framework for spatiotemporal control in the tracking of visual contours
- Pulling Things out of Perspective
- Heteroscedastic Gaussian process regression
- Discrete-Continuous Depth Estimation from a Single Image
- Multiscale conditional random fields for image labeling
- Bayesian SegNet: Model Uncertainty in Deep Convolutional Encoder-Decoder Architectures for Scene Understanding
- Practical Variational Inference for Neural Networks
- A Practical Bayesian Framework for Backpropagation Networks
- Aleatory or epistemic? Does it matter?
- Depth and surface normal estimation from monocular images using regression on deep features and hierarchical CRFs
- Transforming Neural-Net Output Levels to Probability Distributions
- Make3D: Learning 3D Scene Structure from a Single Still Image
- Weight Uncertainty in Neural Network
- Semantic object classes in video: A high-definition ground truth database
Cited by
- Bayesian SegNet: Model Uncertainty in Deep Convolutional Encoder-Decoder Architectures for Scene Understanding
- A survey on deep learning in medical image analysis
- Geometric Loss Functions for Camera Pose Regression with Deep Learning
- Robust and Efficient Transfer Learning with Hidden Parameter Markov Decision Processes
- Multi-task Learning Using Uncertainty to Weigh Losses for Scene Geometry and Semantics
- Opportunities and obstacles for deep learning in biology and medicine
- Bayesian LSTMs in medicine
- On Calibration of Modern Neural Networks
- Uncertainty-Aware Organ Classification for Surgical Data Science Applications in Laparoscopy
- Uncertainty Decomposition in Bayesian Neural Networks with Latent Variables
- 3D Sketching using Multi-View Deep Volumetric Prediction
- Concrete Problems for Autonomous Vehicle Safety: Advantages of Bayesian Deep Learning
- “Dave...I can assure you ...that it’s going to be all right ...” A Definition, Case for, and Survey of Algorithmic Assurances in Human-Autonomy Trust Relationships
- From Deterministic to Generative: Multimodal Stochastic RNNs for Video Captioning
- Uncertainty-Aware Learning from Demonstration Using Mixture Density Networks with Sampling-Free Variance Modeling
- Uncertainties in Parameters Estimated with Neural Networks: Application to Strong Gravitational Lensing
- Deep and Confident Prediction for Time Series at Uber
- Concrete Dropout
- DPC-Net: Deep Pose Correction for Visual Localization
- On the Capacity of Face Representation
Related papers
- Uncertainty analysis and sensitivity analysis for multidisciplinary systems design
- Tolerance analysis approach based on the classification of uncertainty (aleatory / epistemic)
- Bayesian Sensitivity Analysis and Uncertainty Integration for Robust Optimization
- A System Uncertainty Propagation Approach With Model Uncertainty Quantification in Multidisciplinary Design
- Sequential Refinement of Uncertainty Through Bayesian Inference and Global Sensitivity Analysis
- A PDEM-COM framework for uncertainty quantification of backward issues involving both aleatory and epistemic uncertainties
- Evaluation of uncertainty in direct measurement for parameterization of pyrolysis models, Part I: Thermal analysis
- A Calendar Life Assessment of Airborne Products Considering Epistemic and Aleatory Uncertainty