Module kernel_selection 

Kernel selection and cross-validation utilities.

This module provides tools for selecting the best kernel and hyperparameters for a given dataset, including:

• Kernel Target Alignment (KTA): Quick kernel quality metric
• Leave-One-Out Cross-Validation (LOO-CV): Efficient for GP regression
• K-Fold Cross-Validation: For general kernel evaluation
• Kernel comparison utilities: Compare multiple kernels on same data

Example

use tensorlogic_sklears_kernels::kernel_selection::{
    KernelSelector, KernelComparison, KFoldConfig
};
use tensorlogic_sklears_kernels::{RbfKernel, RbfKernelConfig, LinearKernel, Kernel};

// Create kernels to compare
let rbf = RbfKernel::new(RbfKernelConfig::new(0.5)).unwrap();
let linear = LinearKernel::new();

// Sample data
let data = vec![
    vec![1.0, 2.0],
    vec![2.0, 3.0],
    vec![3.0, 4.0],
    vec![4.0, 5.0],
];
let targets = vec![1.0, 2.0, 3.0, 4.0];

// Compare using Kernel Target Alignment
let selector = KernelSelector::new();
let rbf_kta = selector.kernel_target_alignment(&rbf, &data, &targets).unwrap();
let linear_kta = selector.kernel_target_alignment(&linear, &data, &targets).unwrap();

Structs

CrossValidationResult

Results from cross-validation.

GammaSearchResult

Result of gamma grid search for RBF kernel.

KFoldConfig

Configuration for K-fold cross-validation.

KernelComparison

Comparison results for multiple kernels.

KernelSelector

Kernel selector for choosing and comparing kernels.