A General Kernel Framework for Non-CND Distance Measures Using |D|-Dimensional Sparse Landmark Embeddings
Proposes the Sparse Landmark Embedding kernel, guaranteeing PSD kernels for arbitrary distances like geodesic and Wasserstein without CND requirements.
The paper introduces the Sparse Landmark Embedding (SLE) kernel, which embeds inputs via compactly supported bump functions at all |D| training points so any standard PSD kernel applies, removing the Hilbertian (CND) distance requirement that fails on manifolds and distribution spaces. Compact support controls sparsity, keeping kernel matrices well-conditioned despite high dimensionality. The authors prove PSD, sparsity, stability, and universal approximation guarantees, and show SLE matches or exceeds domain-specific baselines using geodesic and Wasserstein distances on accuracy and uncertainty quantification.
- Removes CND distance requirement for Gaussian Process kernels
- Bump-function embeddings yield provably PSD kernels for arbitrary distances
- Validated on geodesic and Wasserstein distances against domain-specific baselines
Full article152 words · extracted from arxiv.org · click to collapse
Kernel methods, and Gaussian Processes (GPs) in particular, require a Hilbertian distance measure---one whose square is conditionally negative definite (CND)---to guarantee positive semi-definiteness (PSD) of the kernel matrix; a condition that fails for many natural input spaces, including smooth manifolds and spaces of probability distributions. We propose the Sparse Landmark Embedding (SLE) kernel, which eliminates this requirement entirely. Each input is embedded into a sparse feature vector via compactly supported bump functions centered at all |D| training points; applying any standard PSD kernel in this embedding space yields a kernel that is provably PSD for arbitrary distance measures. The compact support automatically controls embedding sparsity, keeping kernel matrices well-conditioned and computationally tractable despite the high ambient dimension. We provide theoretical guarantees on PSD, sparsity, stability, and universal approximation, and demonstrate, using geodesic and Wasserstein distances, that the SLE kernel matches or substantially exceeds domain-specific baselines in both predictive accuracy and uncertainty quantification.
Text extracted automatically; images, tables and formatting may be missing. Original: https://arxiv.org/abs/2609.19083