Struct BayesianOptimization 

pub struct BayesianOptimization { /* private fields */ }

Bayesian Optimization for hyperparameter tuning.

Uses Gaussian Processes to model the objective function and acquisition functions to intelligently select the next hyperparameters to evaluate.

Algorithm

1. Initialize with random samples
2. Fit Gaussian Process to observed data
3. Optimize acquisition function to find next point
4. Evaluate objective at new point
5. Repeat steps 2-4 until budget exhausted

Example

use tensorlogic_train::*;
use std::collections::HashMap;

let mut param_space = HashMap::new();
param_space.insert(
    "lr".to_string(),
    HyperparamSpace::log_uniform(1e-4, 1e-1).unwrap(),
);

let mut bayes_opt = BayesianOptimization::new(
    param_space,
    10,  // n_iterations
    5,   // n_initial_points
    42,  // seed
);

// In practice, you would evaluate your model here
// bayes_opt.add_result(result);

Implementations

impl BayesianOptimization

pub fn new( param_space: HashMap<String, HyperparamSpace>, n_iterations: usize, n_initial_points: usize, seed: u64, ) -> Self

Create a new Bayesian Optimization instance.

Arguments

• param_space - Hyperparameter space definition
• n_iterations - Number of optimization iterations
• n_initial_points - Number of random initialization points
• seed - Random seed for reproducibility

pub fn with_acquisition(self, acquisition_fn: AcquisitionFunction) -> Self

Set acquisition function.

pub fn with_kernel(self, kernel: GpKernel) -> Self

Set kernel.

pub fn with_noise(self, noise_variance: f64) -> Self

Set noise variance.

pub fn suggest(&mut self) -> TrainResult<HyperparamConfig>

Suggest next hyperparameter configuration to evaluate.

pub fn add_result(&mut self, result: HyperparamResult)

Add a result from evaluating a configuration.

pub fn best_result(&self) -> Option<&HyperparamResult>

Get the best result found so far.

pub fn sorted_results(&self) -> Vec<&HyperparamResult>

Get all results sorted by score (descending).

pub fn results(&self) -> &[HyperparamResult]

Get all results.

pub fn is_complete(&self) -> bool

Check if optimization is complete.

pub fn current_iteration(&self) -> usize

Get current iteration number.

pub fn total_budget(&self) -> usize

Get total budget (initial + iterations).