Skip to content

scikit-optimize Acquisition Functions

The scikit-optimize (skopt) backend in ALchemist provides a range of acquisition functions for Bayesian optimization using the scikit-optimize library. This guide explains the available options, how to use them, and what each setting means.

Overview

The Acquisition panel in ALchemist allows you to:

  • Choose from several acquisition functions, each balancing exploration and exploitation in different ways.
  • Customize parameters such as exploration/exploitation trade-offs.
  • Run the selected strategy to suggest the next experiment based on your trained model.

Important Note

You must first train your model using the scikit-optimize backend before running any skopt acquisition functions.
See scikit-optimize Backend for details on model training.

Acquisition Functions

  • Expected Improvement (EI):
    Balances exploration and exploitation by selecting points with the highest expected improvement over the current best value.
    Parameter: ξ (xi) — higher values favor exploration.

  • Upper Confidence Bound (UCB):
    Selects points with the highest upper confidence bound, balancing exploration and exploitation.
    Parameter: κ (kappa) — higher values increase exploration.

  • Probability of Improvement (PI):
    Selects points with the highest probability of improving over the current best value.
    Parameter: ξ (xi) — higher values favor exploration.

  • GP Hedge (Auto-balance):
    Automatically balances between EI, UCB, and PI by adaptively selecting the best-performing strategy during optimization.
    Parameters: ξ (xi) and κ (kappa).

Customization:
- Choose to maximize or minimize your objective. - Adjust ξ (xi) and κ (kappa) parameters using sliders as appropriate for the selected acquisition function.

How to Use

  1. Train Model:
    Train your model using the scikit-optimize backend. See scikit-optimize Backend for instructions.

  2. Open Acquisition Panel:
    Go to the Acquisition panel. The scikit-optimize options will appear automatically.

  3. Select Acquisition Function:
    Use the dropdown menu to select the acquisition function (EI, UCB, PI, or GP Hedge).

  4. Configure Options:

  5. Adjust ξ (xi) and κ (kappa) parameters as needed.

  6. Choose whether to maximize or minimize.

  7. Run Acquisition:
    Click Run Acquisition Strategy to suggest the next experiment. Results, including predicted value and uncertainty, will be shown in a notification window and highlighted in the data table and plots.

Model Optimum Finder

In addition to acquisition functions, you can use the Model Prediction Optimum tool to find the point where the model predicts the best value (maximum or minimum).
Note: This does not balance exploration and exploitation—it simply finds the model's optimum prediction.

Tips & Notes

  • Parameter Tuning:
  • Increase ξ (xi) or κ (kappa) for more exploration; decrease for more exploitation.
  • GP Hedge is useful if you are unsure which acquisition function to use.
  • Publication Quality:
    All results and suggested points are integrated with ALchemist's visualization tools for easy analysis and export.

For more details on the underlying algorithms,