methods
TCGPR
What it does
TCGPR studies data consistency through kernel-based Gaussian process modeling, supporting pattern recognition and outlier identification in small materials datasets.
Approach & capabilities
Uses the data sensitivity of kernel-based Gaussian process models.
Defines a consistency measure for pattern recognition and outlier identification.
Explores materials modeling with small datasets and heterogeneous data distributions.
Read more
I proposed TCGPR in 2022, based on the data sensitivity reflected in kernel-based Gaussian process models. It defines a factor to evaluate the data consistency for pattern recognition and outlier identification (https://github.com/Bin-Cao/TCGPR)
This model achieved great performance in studying materials with small data sets. By characterizing the data distributions, we can often achieve better fitting results (though it may not always work).
Following this strategy, we successfully applied the algorithm to two works:(Small, 2024 : https://onlinelibrary.wiley.com/doi/10.1002/smll.202408750) (npj cm 2023 :https://www.nature.com/articles/s41524-023-01150-0).