A Review of Uncertainty Quantification in Deep Learning: Techniques, Applications and Challenges

By Moloud Abdar, Farhad Pourpanah, Sadiq Hussain, Dana Rezazadegan, Li Liu, Mohammad Ghavamzadeh, Paul W. Fieguth, Xiaochun Cao, Abbas Khosravi, U. Rajendra Acharya and 2 more. Information Fusion, 2020

2,678 citations recorded by the source corpus.

Uncertainty quantification (UQ) plays a pivotal role in reduction of uncertainties during both optimization and decision making processes. It can be applied to solve a variety of real-world applications in science and engineering. Bayesian approximation and ensemble learning techniques are two most widely-used UQ methods in the literature. In this regard, researchers have proposed different UQ methods and examined their performance in a variety of applications such as computer vision (e.g., self-driving cars and object detection), image processing (e.g., image restoration), medical image analysis (e.g., medical image classification and segmentation), natural language processing (e.g., text classification, social media texts and recidivism risk-scoring), bioinformatics, etc.This study reviews recent advances in UQ methods used in deep learning. Moreover, we also investigate the application of these methods in reinforcement learning (RL). Then, we outline a few important applications of UQ methods. Finally, we briefly highlight the fundamental research challenges faced by UQ methods and discuss the future research directions in this field.

Subfields: Artificial Intelligence

View Paper: https://www.semanticscholar.org/paper/f14fc9e399d44463a17cc47a9b339b58f6ef7502

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

Citations in the corpus

A Survey on Graph Neural Networks for Time Series: Forecasting, Classification, Imputation, and Anomaly Detection (2023, background): “Uncertainty quantification provides a probabilistic measure of the confidence in the predictions made by the model, aiding in the understanding of the range and likelihood of potential outcomes [300].”
Mutual consistency learning for semi-supervised medical image segmentation (2021, background): “The uncertainty analysis attracts much attention in the both ﬁelds of machine learning and computer vision (Abdar et al., 2021; Jungo and Reyes, 2019).”
Bias in medical AI: Implications for clinical decision-making (2024, methodology): “Quantification of label and prediction uncertainty provides another potential solution, where Bayesian approaches and ensemble methods can be employed to provide confidence intervals or probability distributions of predictions [86–88].”
Self-adaptive loss balanced Physics-informed neural networks (2021, methodology): “Variational inference [29] was used to estimate the posterior distributions of the reconstructed flow [30].”
Remote patient monitoring using artificial intelligence: Current state, applications, and challenges (2023, background): “Abdar et al. (2021) surveyed the research community's work on quantifying uncertainty in machine learning and deep learning models.”
Generating with Confidence: Uncertainty Quantification for Black-box Large Language Models (2023, background): “…machine learning tasks like classification or regression (Lakshminarayanan et al., 2017; Gal & Ghahramani, 2016; Hern´andez-Lobato & Adams, 2015; Abdar et al., 2021a), it bears specific challenges for NLG and has attracted limited attention until recently (Lin et al., 2022a; Kuhn et al., 2023;…”
A survey of uncertainty in deep neural networks (2021, background): “[58] presented an overview on uncertainty quantification methodologies in neural networks and provide an extensive list of references for different application fields and a discussion of open challenges.”
Emotion recognition and artificial intelligence: A systematic review (2014-2023) and research recommendations (2023, methodology): “Some well-known techniques used for UQ involve Bayesian inference, variance-based methods, Monte Carlo methods, probabilistic collocation, ensemble modeling, and bootstrapping [194,198].”
Recent advances and applications of deep learning methods in materials science (2021, background): “While most of the focus in the DL field currently goes into developing new algorithms or training networks to high accuracy, there is increasing attention to UQ, as exemplified by the detailed review of Abdar et al.(285).”
A Review of Generalized Zero-Shot Learning Methods (2020, background): “, COVID-19, or not covered during training but appear in the test phase [3], [4], [5], [6].”