Application of Gaussian processes and transfer learning to prediction and analysis of polymer properties

Revolutionizing Electromagnetic Materials: Machine Learning Enabled Optimization of Polymer Nanocomposites for Enhanced Performance

“…This finding raises the question of which Fe 2 O 3 nanoparticles content should be employed in a PNC to attain a robust RL min value with an optimal RL ≤ À10 dB bandwidth at the minimum thickness (mm). To address this inquiry, we leveraged machine learning, viz., GPR, to model [38][39][40][41] the permittivity and permeability data based on the four tested Fe 2 O 3 nanoparticles datasets (Figure 2A-D) and predicted the most potent Fe 2 O 3 nanoparticles loading in the PDMS as the most potent microwave absorber. GPR, an advanced statistical method rooted in Bayesian learning, employs a parameterfree kernel approach that enables accurate predictions even with limited data.…”

“…GPR, an advanced statistical method rooted in Bayesian learning, employs a parameterfree kernel approach that enables accurate predictions even with limited data. [38][39][40][41] Through the modeling of functions, GPR is capable of generating nonparametric models. [38][39][40][41] The GPR relies upon the covariance function to capture the multivariate Gaussian distribution that is fitted to the data points (Section S1.1, Supporting Information).…”

“…[38][39][40][41] Through the modeling of functions, GPR is capable of generating nonparametric models. [38][39][40][41] The GPR relies upon the covariance function to capture the multivariate Gaussian distribution that is fitted to the data points (Section S1.1, Supporting Information). MATLAB 2022b (R2022b) was used to carry out the GPR predictions.…”

“…The complex permittivity ðε r ¼ ε 0 À ε 00 Þ and permeability ðμ r ¼ μ 0 À μ 00 Þ were then calculated using the standard Nicholson-Ross-Weir method based on the measured S parameters. [44] From the measured ε r and μ r value, the RL is calculated as follows [20][21][22][23][24][25][26][27][28][29][30][31][32][33][34][35][36][37][38] :…”

Recovery of Brine Resources Through Crown-Passivated Graphene, Silicene, and Boron Nitride Nanosheets Based on Machine-Learning Structural Predictions

“…Meanwhile, with the recent integration of ML through the concept of AI in materials science and drug discovery, insightful structural predictions on the role of the CEs of various pore sizes on resource recovery from brine using 2D materials can be made. For instance, Chen et al explored the experimental simulation and Gaussian process regression (GPR) using a small set of data to estimate the polymers and its interaction such as diffusion coefficients and other mechanical properties. The results indicated the superiority of GPR and transfer learning.…”