Robust prediction interval estimation for Gaussian processes by cross-validation method

Naoufal, Acharki,; Bertoncello, Antoine; Garnier, Josselin

doi:10.1016/j.csda.2022.107597

Cited by 11 publications

(9 citation statements)

References 34 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…More generally, the GP predictive distribution may not accurately cover unobserved data. Based on this statement, Acharki et al [45] were interested in correcting the GP estimated hyperparameters to adjust the GP prediction intervals and ensure better coverage probabilities of the GP predictive distribution. In the following, we outline this approach, which is further detailed in Appendix A, before explaining the limits of the method in our application context.…”

Section: Recent Developments For More Reliable Gaussian Process Predi...mentioning

confidence: 99%

“…First of all, Acharki et al [45] assume that he has a first set of estimated GP hyperparameters obtained by MLE or CV method and denoted ( σ 2 0 , θ 0 ). Then, for a given level α of prediction intervals, they consider what they refer to as "Leave-One-Out Coverage Probability" and which corresponds to the ∆(α) given by Eq.…”

Section: A Corrective Approach To Directly Modify the Bounds Of Gauss...mentioning

confidence: 99%

“…The reliability of prediction intervals is rarely considered (except in Petit et al [43]), while it is often affected by poor estimation of GP hyperparameters (as highlighted by Demay et al [44]). To circumvent this problem, other authors such as Acharki et al [45] propose to correct the hyperparameter estimates to obtain more reliable prediction intervals. Whatever the estimation method or correction considered, this calls for validation indicators to control the performance and robustness of GP regression.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Probabilistic surrogate modeling by Gaussian process: A review on recent insights in estimation and validation

Marrel,

Iooss

2024

Reliability Engineering & System Safety

View full text Add to dashboard Cite

Section: Recent Developments For More Reliable Gaussian Process Predi...mentioning

confidence: 99%

Section: A Corrective Approach To Directly Modify the Bounds Of Gauss...mentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Probabilistic surrogate modeling by Gaussian process: A review on recent insights in estimation and validation

Marrel,

Iooss

2024

Reliability Engineering & System Safety

View full text Add to dashboard Cite

“…The radial basis kernel function (Gaussian kernel) was used [21], with the radial basis function (RBF) as an expression for K (x, y) = exp{−∥x − y∥ 2 /2σ 2 }, where x represents the input variable, y is mines the receptive field of the neuron and can affect the model's fitting ability and generalization ability. The penalty coefficient c affects the complexity and accuracy of the model.…”

Section: Model Building 231 Least Squares Support Vector Machine (Lssvm)mentioning

confidence: 99%

Model for predicting the angles of upper limb joints in combination with sEMG and posture capture

Wang,

Xiang,

Zhi

et al. 2023

Meas. Sci. Technol.

View full text Add to dashboard Cite

Since poor man–machine interaction and insufficient coupling occur in the processes of angle prediction and rehabilitation training based purely on the surface electromyography (sEMG) signal, a model for predicting the angles of upper limb joints was presented and validated by experiments. The sEMG and posture capture features were combined to build a hybrid vector, and the intentions of upper limb movements were characterized. The original signals were pre-treated with debiasing, filtering, and noise reduction, and then they were integrated to obtain signal characteristics. Then, feature values in the time domain, frequency domain, time-frequency domain, and entropy were extracted from the treated signals. The snake optimizer least squares support vector machine (SO-LSSVM) was modeled to predict the angles of upper limb joints to improve the poor precision and slow velocity of existing models in the movement control field. Experimental results showed that the prediction model performed well in predicting the motion trails of human upper limb joints from the sEMG signal and attitude information. It effectively reduced both skewing and error in prediction. Hence, it holds great promise for improving the man–machine coupling precision and velocity. Compared to the conventional LSSVM model, the proposed SO-LSSVM model reduced the training time, execution time, and root mean square error of evaluation parameters by 65%, 11%, and 76%, respectively. In summary, the proposed SO-LSSVM model satisfied the real-time requirement for rehabilitation robots and showed high accuracy and robustness.

show abstract

“…4-6) коефіцієнт інжекції залежить від випадкових значень тиску на вході та виході струминного насосу, глибини установки струминного насосу в свердловині, а також від геометричних параметрів сопла, дифузора та інших параметрів. Оскільки центральна гранична теорема теорії ймовірності стверджує, що нормальний розподіл виникає тоді, коли дана випадкова величина являє собою суму великого числа незалежних випадкових величин, кожна з яких відіграє незначну роль в утворенні всієї суми то цілком обґрунтовується розподіл u p , як випадкової величини через нормальний розподіл Гауса [5,6]. Оскільки, за наведеними розрахунками, зміна коефіцієнта інжекції, як випадкової величини допустима в межах від u p max =0,96 до u p min =0 припускаючи, що в більшості випадків значення коефіцієнта інжекції стабілізується в середині інтервалу, а його крайні значення малоймовірні і відповідають правилу 3σ.…”

Section: рис 3 характеристика тиску дроселювання в залежності від под...unclassified

Statistical characteristics of liquid injection by jet pump in the process of hydrocarbon production

Slidenko,

Listovshchyk,

Siryk

2023

Coll. res. pap. nat. min. univ.

View full text Add to dashboard Cite

Purpose.To determine the effective range of pressure ratios in the working fluid at the inlet to the jet pump and at its outlet and to determine the numerical characteristics of the random variable – the injection coefficient, provided that the normal distribution law is implemented and rational values of the injection coefficient are achieved, with the determination of the probability of its implementation. The methods. The calculation is carried out using the statistical numerical characteristics of the injection coefficient and the justification of the initial conditions regarding the ratio of pressures at the inlet and outlet of the jet pump. Findings. The operating pressure range is set within (60...160) bar, which corresponds to the throttling pressure when the liquid passes through the nozzle of the jet pump within the limits accepted in the oil industry. The effect on the value of the injection coefficient separately of the pressure at the inlet and outlet of the jet pump, as well as the complex effect with the determination of the probable range of pressure at the inlet within 360...460 bar and at the outlet 330...336 bar, while -the injection ratio varies from 0 to a maximum value of 0.96. The range of variation of the injection coefficient within 3 σ with the centering of the random variable and the determination of the mathematical expectation with the value of 0.48 is constructed. The originality. It was found that within half of the mean square deviation, the established value of the probability of realization of the injection coefficient is 0.383. Within the mean square deviation, the established value of the probability of realizing the injection coefficient is 0.683, which confirms the high probability of realizing the injection coefficient with a centering of 0.48 Practical implementation. The obtained results of the calculations make it possible to substantiate the design parameters and energy parameters of the system of supplying the working fluid from the surface during the design of the pump and the development of the technology of its application with the achievement of rational values of the injection coefficient. This will increase the efficiency of the jet pump for cleaning the near-stem area of the oil well and will contribute to the increase of hydrocarbon production.

show abstract

Robust prediction interval estimation for Gaussian processes by cross-validation method

Cited by 11 publications

References 34 publications

Probabilistic surrogate modeling by Gaussian process: A review on recent insights in estimation and validation

Probabilistic surrogate modeling by Gaussian process: A review on recent insights in estimation and validation

Model for predicting the angles of upper limb joints in combination with sEMG and posture capture

Statistical characteristics of liquid injection by jet pump in the process of hydrocarbon production

Contact Info

Product

Resources

About