SEARCH · Engineering Papers
Results for “regression”
Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.
Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.
Assessments of epistemic uncertainty using Gaussian stochastic weight averaging for fluid-flow regression
Here, we use Gaussian stochastic weight averaging (SWAG) to assess the epistemic uncertainty associated with neural-network-based function approximation relevant to fluid flows. SWAG approximates a posterior Gaussian distribution of each weight, given training data, and a constant learning rate. Having access to this distribution, it is able to create multiple models with various combinations of sampled weights, which can be used to obtain ensemble predictions. The average of such an ensemble can be regarded as the 'mean estimation', whereas its standard deviation can be used to construct 'confidence intervals', which enable us to perform uncertainty quantification (UQ) with regard to the training process of neural networks. We utilize representative neural-network-based function approximation tasks for the following cases: (i) a two-dimensional circular-cylinder wake; (ii) the DayMET dataset (maximum daily temperature in North America); (iii) a three-dimensional square-cylinder wake; and (iv) urban flow, to assess the generalizability of the present idea for a wide range of complex datasets. SWAG-based UQ can be applied regardless of the network architecture, and therefore, we demonstrate the applicability of the method for two types of neural networks: (i) global field reconstruction from sparse sensors by combining convolutional neural network (CNN) and multi-layer perceptron (MLP); and (ii) far-field state estimation from sectional data with two-dimensional CNN. We find that SWAG can obtain physically-interpretable confidence-interval estimates from the perspective of epistemic uncertainty. This capability supports its use for a wide range of problems in science and engineering.
Bayesian linear regression for surface roughness prediction
Not Available
Pairwise Difference Regression: A Machine Learning Meta-algorithm for Improved Prediction and Uncertainty Quantification in Chemical Search
Not Available
Lattice Thermal Conductivity Prediction Using Symbolic Regression and Machine Learning
Not provided.
Pursuit of the Ultimate Regression Model for Samarium(III), Europium(III), and LiCl Using Laser-Induced Fluorescence, Design of Experiments, and a Genetic Algorithm for Feature Selection
Not Available
Optical Turbulence Profile Modeling in the Atmospheric Boundary Layer: A Random Forest Regression Approach
Explore the source record for details and available documents.
A selective small-molecule STAT5 PROTAC degrader capable of achieving tumor regression in vivo
Not provided.
Supervised machine learning-based multivariate regression of parallel closures for a high-collisionality deuterium-carbon plasma
Many plasmas of interest in laboratory experiments and space consist of multiple ion species. In tokamak edge plasmas, for instance, ionized impurities expelled from the vessel wall influence plasma transport. When describing multi-species plasmas using fluid equations, we need accurate closure relations to close the set of fluid equations. In this study, we introduce the development of fitting formulas for parallel closures using supervised machine learning, in conjunction with the recent closure theory, considering multi-ion collisions and arbitrary ion temperatures. We apply this approach to a high-collisionality deuterium-carbon plasma and demonstrate its effectiveness. As a result, the machine learning-based method for developing practical and accurate closures can be extended to a wider range of plasmas.
Partial molar properties from molecular simulation using multiple linear regression
Partial molar volumes, energies, and enthalpies can be computed from NpT-Gibbs ensemble simulations through a post-processing procedure that leverages fluctuations in composition, total volume, and total energy of a simulation box. By recording the instantaneous box volumes V and instantaneous number of molecules N i of each of n species for M frames, a large M ×n matrix N is constructed, as well as the M × 1 vector V. The 1 × n vector of partial molar volumes V¯ may then be solved using N • V¯=V. A similar construction permits calculation of partial molar energies using M instantaneous measurements of the total energy of the simulation box, and N • U¯=U. Partial molar enthalpies may be derived from U¯, V¯, and pressure p. These properties may be used to calculate enthalpy and entropy of transfer (absorption, extraction, and adsorption) for species in complex mixtures. The method is demonstrated on three systems in the NpT-Gibbs ensemble: a highly compressible natural gas condensate of methane, n-butane, and n-decane, the liquid-phase adsorption of 1,5-pentanediol and ethanol onto the MFI zeolite, and a relatively incompressible mixture of ethanol, n-dodecane, and water at liquid-liquid equilibrium. Property predictions are compared to those from numerical differentiation of simulation data sequentially changing the composition and from equations of state. Here, the method can also be extended to reaction equilibria in a closed system and is applied to a reactive first-principles Monte Carlo simulation of compressed nitrogen/oxygen.
Robust estimation and outlier detection for varying-coefficient models via penalized regression
Not provided.