The Hidden Uncertainty in a Neural Networks Activations
- URL: http://arxiv.org/abs/2012.03082v2
- Date: Tue, 23 Feb 2021 08:43:00 GMT
- Title: The Hidden Uncertainty in a Neural Networks Activations
- Authors: Janis Postels, Hermann Blum, Yannick Str\"umpler, Cesar Cadena, Roland
Siegwart, Luc Van Gool, Federico Tombari
- Abstract summary: The distribution of a neural network's latent representations has been successfully used to detect out-of-distribution (OOD) data.
This work investigates whether this distribution correlates with a model's epistemic uncertainty, thus indicating its ability to generalise to novel inputs.
- Score: 105.4223982696279
- License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
- Abstract: The distribution of a neural network's latent representations has been
successfully used to detect out-of-distribution (OOD) data. This work
investigates whether this distribution moreover correlates with a model's
epistemic uncertainty, thus indicates its ability to generalise to novel
inputs. We first empirically verify that epistemic uncertainty can be
identified with the surprise, thus the negative log-likelihood, of observing a
particular latent representation. Moreover, we demonstrate that the
output-conditional distribution of hidden representations also allows
quantifying aleatoric uncertainty via the entropy of the predictive
distribution. 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 more
expensive - methods (e.g. deep ensembles). While our approach does not require
modifying the training process, we follow prior work and experiment with an
additional regularising loss that increases the information in the latent
representations. We find that this leads to improved OOD detection of epistemic
uncertainty at the cost of ambiguous calibration close to the data
distribution. We verify our findings on both classification and regression
models.
Related papers
- The Epistemic Uncertainty Hole: an issue of Bayesian Neural Networks [0.6906005491572401]
We show that the evolution of "epistemic uncertainty metrics" regarding the model size and the size of the training set, goes against theoretical expectations.
This phenomenon, which we call "epistemic uncertainty hole", is all the more problematic as it undermines the entire applicative potential of BDL.
arXiv Detail & Related papers (2024-07-02T06:54:46Z) - Identifiable Latent Neural Causal Models [82.14087963690561]
Causal representation learning seeks to uncover latent, high-level causal representations from low-level observed data.
We determine the types of distribution shifts that do contribute to the identifiability of causal representations.
We translate our findings into a practical algorithm, allowing for the acquisition of reliable latent causal representations.
arXiv Detail & Related papers (2024-03-23T04:13:55Z) - Learning Linear Causal Representations from Interventions under General
Nonlinear Mixing [52.66151568785088]
We prove strong identifiability results given unknown single-node interventions without access to the intervention targets.
This is the first instance of causal identifiability from non-paired interventions for deep neural network embeddings.
arXiv Detail & Related papers (2023-06-04T02:32:12Z) - Looking at the posterior: accuracy and uncertainty of neural-network
predictions [0.0]
We show that prediction accuracy depends on both epistemic and aleatoric uncertainty.
We introduce a novel acquisition function that outperforms common uncertainty-based methods.
arXiv Detail & Related papers (2022-11-26T16:13:32Z) - Robust uncertainty estimates with out-of-distribution pseudo-inputs
training [0.0]
We propose to explicitly train the uncertainty predictor where we are not given data to make it reliable.
As one cannot train without data, we provide mechanisms for generating pseudo-inputs in informative low-density regions of the input space.
With a holistic evaluation, we demonstrate that this yields robust and interpretable predictions of uncertainty while retaining state-of-the-art performance on diverse tasks.
arXiv Detail & Related papers (2022-01-15T17:15:07Z) - Dense Uncertainty Estimation via an Ensemble-based Conditional Latent
Variable Model [68.34559610536614]
We argue that the aleatoric uncertainty is an inherent attribute of the data and can only be correctly estimated with an unbiased oracle model.
We propose a new sampling and selection strategy at train time to approximate the oracle model for aleatoric uncertainty estimation.
Our results show that our solution achieves both accurate deterministic results and reliable uncertainty estimation.
arXiv Detail & Related papers (2021-11-22T08:54:10Z) - Aleatoric uncertainty for Errors-in-Variables models in deep regression [0.48733623015338234]
We show how the concept of Errors-in-Variables can be used in Bayesian deep regression.
We discuss the approach along various simulated and real examples.
arXiv Detail & Related papers (2021-05-19T12:37:02Z) - Improving Uncertainty Calibration via Prior Augmented Data [56.88185136509654]
Neural networks have proven successful at learning from complex data distributions by acting as universal function approximators.
They are often overconfident in their predictions, which leads to inaccurate and miscalibrated probabilistic predictions.
We propose a solution by seeking out regions of feature space where the model is unjustifiably overconfident, and conditionally raising the entropy of those predictions towards that of the prior distribution of the labels.
arXiv Detail & Related papers (2021-02-22T07:02:37Z) - DEUP: Direct Epistemic Uncertainty Prediction [56.087230230128185]
Epistemic uncertainty is part of out-of-sample prediction error due to the lack of knowledge of the learner.
We propose a principled approach for directly estimating epistemic uncertainty by learning to predict generalization error and subtracting an estimate of aleatoric uncertainty.
arXiv Detail & Related papers (2021-02-16T23:50:35Z) - Posterior Network: Uncertainty Estimation without OOD Samples via
Density-Based Pseudo-Counts [33.45069308137142]
Posterior Network (PostNet) predicts an individual closed-form posterior distribution over predicted probabilites for any input sample.
PostNet achieves state-of-the art results in OOD detection and in uncertainty calibration under dataset shifts.
arXiv Detail & Related papers (2020-06-16T15:16:32Z)
This list is automatically generated from the titles and abstracts of the papers in this site.
This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.