Journal: Earth's Future

Article Title: The Nonlinear and Distinct Responses of Ocean Heat Content and Anthropogenic Carbon to Ice Sheet Freshwater Discharge in a Warming Climate

doi: 10.1029/2024EF004475

Figure Lengend Snippet: (a) Normalized Root Mean Square Error (nRMSE) values between the Gaussian Predictor Response (GPR) predicted OHC ANTH and actual OHC ANTH generated by withholding one predictor at a time for the Control (gray), AIS (blue), GrIS (pink), and AGrIS (black) simulations. (b) Same as for panel (a) but for C ANTH .

Article Snippet: Here, we identify the different driving factors in the FW linear and nonlinear OHC ANTH and C ANTH responses using a predictive, Gaussian Process Regression (GPR) model in MATLAB's Regression Learner toolbox.

Techniques: Generated, Control

Journal: Scientific Reports

Article Title: Machine learning based on computational fluid dynamics enables geometric design optimisation of the NeoVAD blades

doi: 10.1038/s41598-023-33708-9

Figure Lengend Snippet: Regression analysis of baseline Multilinear Regression and Machine Learning methods—Gaussian Process Regression and Bayesian Regularised Artificial Neural Network. Output along the y -axis shows the normalised output of the surrogate at each of the 32 designs and Target along the x -axis shows normalised simulations results for efficiency at these design points.

Article Snippet: A Gaussian Process Regression model was implemented in MATLAB using a five-fold cross-validation method whereby the data is partitioned to exclude a fifth of available set for training a validation.

Techniques:

Journal: Scientific Reports

Article Title: Machine learning based on computational fluid dynamics enables geometric design optimisation of the NeoVAD blades

doi: 10.1038/s41598-023-33708-9

Figure Lengend Snippet: Comparison of previously best performing pump design from the original 32 base designs and the new optimised blade design at selected operating point of Q = 2 L/min, H = 70 mmHg—the result of the optimisation routine utilising constraint iteration 3 and the Gaussian process regression surrogate model.

Article Snippet: A Gaussian Process Regression model was implemented in MATLAB using a five-fold cross-validation method whereby the data is partitioned to exclude a fifth of available set for training a validation.

Techniques: