Module kernel 

The DeepKernel type — a Deep Kernel Learning wrapper that composes a base kernel with a neural feature extractor.

Given a base kernel K_base and a feature map g_θ, the Deep Kernel is

K_DKL(x, y) = K_base(g_θ(x), g_θ(y)).

This generic wrapper implements the crate-level Kernel trait so a DeepKernel can slot into any downstream machinery that consumes dyn Kernel (SVM adapters, Gram-matrix utilities, kernel-alignment search, etc.).

The base kernel and feature extractor are both owned by the wrapper. Cloning clones both; mutating parameters requires holding a &mut DeepKernel and going through DeepKernel::feature_extractor_mut.

Structs

DeepKernel

Composition of a neural feature extractor with a classical kernel.

DeepKernelSummary

Debug helper — prints extractor shape and base kernel name.

Traits

FeatureMapShape

Helper trait for kinds of feature extractor whose output dimension matches the base kernel’s expected input dimension. Implemented automatically for every NeuralFeatureMap; exists purely as a documentation anchor.