Skip to main content

Simulations, Data Analysis and Algorithms

Computational Astrophysics and Cosmology Cover Image

TableĀ 5 Performance of the classification methods in this work. The accuracy is reported for the binary, sequential binary, and multiclass classification models. For the sequential binary and multiclass classifiers, the labels are analogous to the number of post-impact remnants (0,1,2). For the binary classifiers, the labels correspond to whether the LR/SLR exists (1) or not (0)

From: Machine learning applied to simulations of collisions between rotating, differentiated planets

Type Classes Method Accuracy
Binary (LR) 0 | 1,2 MLP 0.9879
XGB 0.9882
Binary (SLR) 0,1 | 2 MLP 0.9680
XGB 0.9731
Sequential Binary 0 | 1 | 2 MLP 0.9563
XGB 0.9616
Multiclass 0 | 1 | 2 MLP 0.9532
XGB 0.9627
\