The code in this repository is intended to replicate Figure 4 in the paper "Efficient Multiscale Gaussian Process Regression Using Hierarchial Clustering." Questions and concerns about the code should be directed towards Ze Jia Zhang (zzejia at umich dot edu).
- MGPR.pdf: preprint PDF version of the paper.
- Figure4.m: script to produce the desired figure.
- Train_Kern_Std.m: training algorithm for standard GPR.
- Test_Kern_Std.m: testing algorithm for standard GPR.
- Train_fd_Multiscale_F1i.m: training algorithm for MGPR.
- Test_fd_Multiscale_F1c.m: testing algorithm for MGPR.
- optLML_Multiscale2.m: optimization container for hyperparameters.
- hcluster0.m: clustering algorithm for MGPR.
- GaussMx.m, GaussMxnd.m: construction of Gaussian kernels.
- Dist2.m: compute pairwise distance between points.