A review of uncertainty quantification in deep learning: Techniques, applications and challenges

Abdar, Moloud; Pourpanah, Farhad; Hussain, Sadiq; Rezazadegan, Dana; Liu, Li; Ghavamzadeh, Mohammad; Fieguth, Paul; Cao, Xiaochun; Khosravi, Abbas; Acharya, U. Rajendra; Makarenkov, Vladimir; Nahavandi, Saeid

doi:10.1016/j.inffus.2021.05.008

Cited by 1,071 publications

(453 citation statements)

References 366 publications

Supporting

Mentioning

452

Contrasting

Unclassified

Order By: Relevance

“…There are many tools for estimating uncertainty, such as bootstrapping, quantile regression, Bayesian inference, and dropout for the neural networks [35][36][37]. Depending on different models, one technique may be more suitable than others; therefore, they will be described in more detail in the subsequent sections, which introduce the tested models.…”

Section: Data-driven Models and Uncertainty Estimationmentioning

confidence: 99%

“…One way to mitigate and manage the risk of making dangerous decisions is to estimate the prediction uncertainty. There are a number of methods for estimating it with an NN, such as Bayesian methods and bagging [36], but the most popular is certainly by using dropout [38]. This technique was originally developed to avoid the co-adaptation of the parameters during the training of the network in order to reduce overfitting and improve the generalization error.…”

Section: Feed-forward Neural Networkmentioning

confidence: 99%

See 1 more Smart Citation

Uncertainty Estimation for Machine Learning Models in Multiphase Flow Applications

et al. 2021

View full text Add to dashboard Cite

In oil and gas production, it is essential to monitor some performance indicators that are related to the composition of the extracted mixture, such as the liquid and gas content of the flow. These indicators cannot be directly measured and must be inferred with other measurements by using soft sensor approaches that model the target quantity. For the purpose of production monitoring, point estimation alone is not enough, and a confidence interval is required in order to assess the uncertainty in the provided measure. Decisions based on these estimations can have a large impact on production costs; therefore, providing a quantification of uncertainty can help operators make the most correct choices. This paper focuses on the estimation of the performance indicator called the water-in-liquid ratio by using data-driven tools: firstly, anomaly detection techniques are employed to find data that can alter the performance of the subsequent model; then, different machine learning models, such as Gaussian processes, random forests, linear local forests, and neural networks, are tested and employed to perform uncertainty-aware predictions on data coming from an industrial tool, the multiphase flow meter, which collects multiple signals from the flow mixture. The reported results show the differences between the discussed approaches and the advantages of the uncertainty estimation; in particular, they show that methods such as the Gaussian process and linear local forest are capable of reaching competitive performance in terms of both RMSE (1.9–2.1) and estimated uncertainty (1.6–2.6).

show abstract

Section: Data-driven Models and Uncertainty Estimationmentioning

confidence: 99%

Section: Feed-forward Neural Networkmentioning

confidence: 99%

Uncertainty Estimation for Machine Learning Models in Multiphase Flow Applications

et al. 2021

View full text Add to dashboard Cite

show abstract

“…However, few datasets exist with train and test subsets for applications of machine learning and deep learning methods with UQ in regression 4,5 . UQ is getting vast popularity due to its demand in ML and DL methods [6][7][8][9] . In most of the previous studies, researchers in UQ are splitting the dataset randomly 10,11 .…”

Section: Background and Summarymentioning

confidence: 99%

“…We train networks for point prediction. Then, we compute the upper and lower bounds with a Gaussian and homoscedastic uncertainty assumption 9,28 . In the Gaussian and homoscedastic uncertainty assumption, the PI is presented as [µ − zσ , µ + zσ ].…”

Section: Technical Validation: Model Training For Initial Performancementioning

confidence: 99%

Synthetic Datasets for Numeric Uncertainty Quantification

Kabir

Abdar

Khosravi

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

In this paper, we propose ten synthetic datasets for point prediction and numeric uncertainty quantification (UQ). These datasets are split into train, validation, and test sets for model benchmarking. Equations and the description of each dataset are provided in detail. We also present representative shallow neural network (NN) training and Random Vector Functional Link (RVFL) training examples. Both training train models for the point prediction. We perform uncertainty quantification with the consideration of a gaussian and homoscedastic distribution. As the distribution consideration and models are rudimental, much room exists for further explorations and improvements. The dataset and scripts are available at the following link: https://github.com/dipuk0506/UQ-Data

show abstract

“…In recent years, machine learning algorithms have become a more robust approach in landslide research [43], but the models require managing uncertainties. These uncertainties could result from errors and model variability [44], difficulties in the selection of parameters [45], system understanding [46], the weighting of parameters [47], and human judgment [48]. Moreover, machine learning may encounter prediction errors if trained with a small data set [43].…”

Section: Introductionmentioning

confidence: 99%

Evaluation of the Effect of Hydroseeded Vegetation for Slope Reinforcement

Emeka

Nahazanan

Kalantar

et al. 2021

Land

View full text Add to dashboard Cite

A landslide is a significant environmental hazard that results in an enormous loss of lives and properties. Studies have revealed that rainfall, soil characteristics, and human errors, such as deforestation, are the leading causes of landslides, reducing soil water infiltration and increasing the water runoff of a slope. This paper introduces vegetation establishment as a low-cost, practical measure for slope reinforcement through the ground cover and the root of the vegetation. This study reveals the level of complexity of the terrain with regards to the evaluation of high and low stability areas and has produced a landslide susceptibility map. For this purpose, 12 conditioning factors, namely slope, aspect, elevation, curvature, hill shade, stream power index (SPI), topographic wetness index (TWI), terrain roughness index (TRI), distances to roads, distance to lakes, distance to trees, and build-up, were used through the analytic hierarchy process (AHP) model to produce landslide susceptibility map. Receiver operating characteristics (ROC) was used for validation of the results. The area under the curve (AUC) values obtained from the ROC method for the AHP model was 0.865. Four seed samples, namely ryegrass, rye corn, signal grass, and couch, were hydroseeded to determine the vegetation root and ground cover’s effectiveness on stabilization and reinforcement on a high-risk susceptible 65° slope between August and December 2020. The observed monthly vegetation root of couch grass gave the most acceptable result. With a spreading and creeping vegetation ground cover characteristic, ryegrass showed the most acceptable monthly result for vegetation ground cover effectiveness. The findings suggest that the selection of couch species over other species is justified based on landslide control benefits.

show abstract

A review of uncertainty quantification in deep learning: Techniques, applications and challenges

Cited by 1,071 publications

References 366 publications

Uncertainty Estimation for Machine Learning Models in Multiphase Flow Applications

Uncertainty Estimation for Machine Learning Models in Multiphase Flow Applications

Synthetic Datasets for Numeric Uncertainty Quantification

Evaluation of the Effect of Hydroseeded Vegetation for Slope Reinforcement

Contact Info

Product

Resources

About