Module kernel_pca 

Kernel Principal Component Analysis (Scholkopf, Smola & Muller, 1998).

This module provides a fully self-contained implementation of Kernel PCA that operates through the crate’s Kernel trait. Any kernel in the crate (RBF, Linear, Polynomial, Symbolic, etc.) can be plugged in; the only additional requirement for fitting is Clone + 'static, which every shipped kernel satisfies.

Module map

Submodule	Purpose
centering	Double-centering of the Gram matrix + test-time centering
eigendecomp	Wrapper around scirs2_linalg::eigh for top-k eigenpairs
error	KernelPcaError / KernelPcaResult types
model	KernelPCA, KernelPcaConfig, FittedKernelPCA

Minimal example

use tensorlogic_sklears_kernels::kernel_pca::{KernelPCA, KernelPcaConfig};
use tensorlogic_sklears_kernels::RbfKernel;
use tensorlogic_sklears_kernels::RbfKernelConfig;

let kernel = RbfKernel::new(RbfKernelConfig::new(1.0)).expect("kernel");
let config = KernelPcaConfig::new(2);
let model = KernelPCA::build(kernel, config).expect("model");

let data = vec![
    vec![1.0, 2.0],
    vec![3.0, 4.0],
    vec![5.0, 6.0],
];
let fitted = model.fit(&data).expect("fit");
let embedding = fitted.transform(&data).expect("transform");
assert_eq!(embedding.nrows(), 3);
assert_eq!(embedding.ncols(), 2);

References

• Scholkopf, B., Smola, A. & Muller, K.-R. (1998). Nonlinear Component Analysis as a Kernel Eigenvalue Problem. Neural Computation 10(5), 1299–1319.

Re-exports

pub use centering::center_test_kernel;

pub use centering::double_center;

pub use centering::KernelCenteringStats;

pub use eigendecomp::symmetric_eigendecomp;

pub use eigendecomp::TopKEigen;

pub use error::KernelPcaError;

pub use error::KernelPcaResult;

pub use model::FittedKernelPCA;

pub use model::KernelPCA;

pub use model::KernelPcaConfig;

Modules

centering

Double-centering utilities for Kernel PCA.

eigendecomp

Wrappers around scirs2_linalg::eigh tailored to Kernel PCA.

error

Error type for the Kernel PCA research-preview module.

model

The KernelPCA estimator and its fitted counterpart FittedKernelPCA.