A survey on epistemic (model) uncertainty in supervised learning: Recent advances and applications

Zhou, Xinlei; Liu, Han; Pourpanah, Farhad; Zeng, Tieyong; Wang, Xizhao

doi:10.1016/j.neucom.2021.10.119

Cited by 35 publications

(8 citation statements)

References 90 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Here, we intend to present a complete view of all topics along with their interplay. For instance, loss functions are an essential part of a machine learning algorithm; however, as can be seen from some recent review papers in deep learning (Abdar et al 2021;Zhou et al 2022) it is missing from the literature, which focuses on simulation-based techniques, either in simple simulation settings or in Bayesian ones.…”

Section: Summary Of Review Papers On Probabilistic Forecasting and Pr...mentioning

confidence: 99%

A review of predictive uncertainty estimation with machine learning

Tyralis,

Papacharalampous

2024

Artif Intell Rev

View full text Add to dashboard Cite

Predictions and forecasts of machine learning models should take the form of probability distributions, aiming to increase the quantity of information communicated to end users. Although applications of probabilistic prediction and forecasting with machine learning models in academia and industry are becoming more frequent, related concepts and methods have not been formalized and structured under a holistic view of the entire field. Here, we review the topic of predictive uncertainty estimation with machine learning algorithms, as well as the related metrics (consistent scoring functions and proper scoring rules) for assessing probabilistic predictions. The review covers a time period spanning from the introduction of early statistical (linear regression and time series models, based on Bayesian statistics or quantile regression) to recent machine learning algorithms (including generalized additive models for location, scale and shape, random forests, boosting and deep learning algorithms) that are more flexible by nature. The review of the progress in the field, expedites our understanding on how to develop new algorithms tailored to users’ needs, since the latest advancements are based on some fundamental concepts applied to more complex algorithms. We conclude by classifying the material and discussing challenges that are becoming a hot topic of research.

show abstract

Section: Summary Of Review Papers On Probabilistic Forecasting and Pr...mentioning

confidence: 99%

A review of predictive uncertainty estimation with machine learning

Tyralis,

Papacharalampous

2024

Artif Intell Rev

View full text Add to dashboard Cite

show abstract

“…Supervised learning remains in tasks such as image classification, speech recognition, natural language processing, and many other applications. As more labeled data becomes available and computing power increases, supervised learning techniques can be further refined to improve accuracy and performance (Aminian, Abroshan, Khalili, Toni, & Rodrigues, 2022;Zhou, Liu, Pourpanah, Zeng, & Wang, 2022;Iman, Arabnia, & Rasheed, 2023).…”

Section: Synergy Explanationmentioning

confidence: 99%

Review of synergy between machine learning and first principles models for asset integrity management

Cai,

Fang,

Daida

et al. 2023

Front. Chem. Eng.

View full text Add to dashboard Cite

The chemical process industry (CPI) accumulated a rich data asset through industrial 4.0. There is a strong drive to develop and utilize effective approaches for process performance prediction and improvement, process control, sensor development, asset management, etc. The synergy between machine learning and first principles models can bring new insights and add tremendous value to the CPI. This paper reviews various applications of the synergies towards asset integrity management. An overview of some related commercial software packages are also provided.

show abstract

“…While epistemic uncertainties are subjective. With continuing engineering progress and increasing data information, epistemic uncertainties can be reduced (Zhou et al, 2022;Tang et al, 2017;He et al, 2015;Nannapaneni and Mahadevan, 2016).…”

Section: Introductionmentioning

confidence: 99%

Multidisciplinary design optimization of engineering systems under uncertainty: a review

Meng

Yang

et al. 2022

IJSI

View full text Add to dashboard Cite

PurposeAs an advanced calculation methodology, reliability-based multidisciplinary design optimization (RBMDO) has been widely acknowledged for the design problems of modern complex engineering systems, not only because of the accurate evaluation of the impact of uncertain factors but also the relatively good balance between economy and safety of performance. However, with the increasing complexity of engineering technology, the proposed RBMDO method gradually cannot effectively solve the higher nonlinear coupled multidisciplinary uncertainty design optimization problems, which limits the engineering application of RBMDO. Many valuable works have been done in the RBMDO field in recent decades to tackle the above challenges. This study is to review these studies systematically, highlight the research opportunities and challenges, and attempt to guide future research efforts.Design/methodology/approachThis study presents a comprehensive review of the RBMDO theory, mainly including the reliability analysis methods of different uncertainties and the decoupling strategies of RBMDO.FindingsFirst, the multidisciplinary design optimization (MDO) preliminaries are given. The basic MDO concepts and the corresponding mathematical formulas are illustrated. Then, the procedures of three RBMDO methods with different reliability analysis strategies are introduced in detail. These RBMDO methods were proposed for the design optimization problems under different uncertainty types. Furtherly, an optimization problem for a certain operating condition of a turbine runner blade is introduced to illustrate the engineering application of the above method. Finally, three aspects of future challenges for RBMDO, namely, time-varying uncertainty analysis; high-precision surrogate models, and verification, validation and accreditation (VVA) for the model, are discussed followed by the conclusion.Originality/valueThe scope of this study is to introduce the RBMDO theory systematically. Three commonly used RBMDO-SORA methods are reviewed comprehensively, including the methods' general procedures and mathematical models.

show abstract

A survey on epistemic (model) uncertainty in supervised learning: Recent advances and applications

Cited by 35 publications

References 90 publications

A review of predictive uncertainty estimation with machine learning

A review of predictive uncertainty estimation with machine learning

Review of synergy between machine learning and first principles models for asset integrity management

Multidisciplinary design optimization of engineering systems under uncertainty: a review

Contact Info

Product

Resources

About