Generalized Kernel Regularized Least Squares

• In Event: Computational Methods: New Approaches and Applications

Brief Overview

We generalize KRLS to non-linear models by representing the problem as a hierarchical model. This permits easy estimation and other extensions such as estimating multiple kernels in a single model. We use random sketching to scale to large datasets.

Authors

• Qing Chang, University of Pittsburgh

• Max Harrison Goplerud, University of Pittsburgh