bayes_opt.acquisition.UpperConfidenceBound

class bayes_opt.acquisition.UpperConfidenceBound(kappa: float = 2.576, exploration_decay: float | None = None, exploration_decay_delay: int | None = None, random_state: int | RandomState | None = None) → None

Upper Confidence Bound acquisition function.

The upper confidence bound is calculated as

\[\text{UCB}(x) = \mu(x) + \kappa \sigma(x).\]

Parameters:

• kappa (float) – Governs the exploration/exploitation tradeoff. Lower prefers exploitation, higher prefers exploration.

• exploration_decay (float | None) – Decay rate for kappa. If None, no decay is applied.

• exploration_decay_delay (int | None) – Delay for decay. If None, decay is applied from the start.

• random_state (int | RandomState | None)

base_acq(mean: ndarray[tuple[Any, ...], dtype[floating[Any]]], std: ndarray[tuple[Any, ...], dtype[floating[Any]]]) → ndarray[tuple[Any, ...], dtype[floating[Any]]]

Calculate the upper confidence bound.

Parameters:

• mean (ndarray[tuple[Any, ...], dtype[floating[Any]]]) – Mean of the predictive distribution.

• std (ndarray[tuple[Any, ...], dtype[floating[Any]]]) – Standard deviation of the predictive distribution.

Return type:

ndarray[tuple[Any, ...], dtype[floating[Any]]]

Returns:

np.ndarray – Acquisition function value.

decay_exploration() → None

Decay kappa by a constant rate.

Adjust exploration/exploitation trade-off by reducing kappa.

Note

This method is called automatically at the end of each suggest() call.

Return type:

None

get_acquisition_params() → dict[str, Any]

Get the current acquisition function parameters.

Return type:

dict[str, Any]

Returns:

dict – Dictionary containing the current acquisition function parameters.

set_acquisition_params(params: dict[str, Any]) → None

Set the acquisition function parameters.

Parameters:

params (dict[str, Any]) – Dictionary containing the acquisition function parameters.

Return type:

None

suggest(gp: GaussianProcessRegressor, target_space: TargetSpace, n_random: int = 10000, n_smart: int = 10, fit_gp: bool = True, random_state: int | RandomState | None = None) → ndarray[tuple[Any, ...], dtype[floating[Any]]]

Suggest a promising point to probe next.

Parameters:

• gp (GaussianProcessRegressor) – A fitted Gaussian Process.

• target_space (TargetSpace) – The target space to probe.

• n_random (int) – Number of random samples to use.

• n_smart (int) – Number of starting points for the L-BFGS-B optimizer.

• fit_gp (bool) – Whether to fit the Gaussian Process to the target space. Set to False if the GP is already fitted.

• random_state (int | RandomState | None) – Random state to use for the optimization.

Return type:

ndarray[tuple[Any, ...], dtype[floating[Any]]]

Returns:

np.ndarray – Suggested point to probe next.