Not-So-Random Features
This work addresses kernel learning for machine learning practitioners, offering a scalable and improved alternative to random features methods, though it appears incremental as it builds on existing kernel and SVM frameworks.
The authors tackled the problem of kernel learning by proposing a principled method based on Fourier analysis for translation-invariant or rotation-invariant kernels, which iteratively refines SVM margins and shows consistent improvements over random features-based methods in evaluations on synthetic and real-world datasets.
We propose a principled method for kernel learning, which relies on a Fourier-analytic characterization of translation-invariant or rotation-invariant kernels. Our method produces a sequence of feature maps, iteratively refining the SVM margin. We provide rigorous guarantees for optimality and generalization, interpreting our algorithm as online equilibrium-finding dynamics in a certain two-player min-max game. Evaluations on synthetic and real-world datasets demonstrate scalability and consistent improvements over related random features-based methods.