ALEF

What it does

Availability

The repository page lists no releases or packages.checked 14 Nov 2026

Bayesian optimization

The Bayesian optimization module implements query-based optimization with acquisition functions such as GP-UCB and EI.checked 14 Nov 2026

Design methods

Its experimental-design methods include pool-based, oracle-based, and safe active learning, as well as Bayesian optimization.checked 14 Nov 2026

Examples

The main usage of the repository's building blocks is illustrated in Jupyter notebooks.checked 14 Nov 2026

License

The software is open-sourced under the AGPL-3.0 license.checked 14 Nov 2026

Models and kernels

Implemented models include standard and marginalized-hyperparameter GPs and Deep GPs; kernels include Deep Kernels and Neural Kernel Network.checked 14 Nov 2026

Pool-based learning

The pool-based learner can use acquisition functions including Entropy and Variance and validation metrics including RMSE and NLL.checked 14 Nov 2026

Purpose

ALEF provides classes and scripts for active learning of stochastic models such as Gaussian Processes.checked 14 Nov 2026

Safe learning

Safe active learning uses a safety model, a data pool with safety labels, and a safety threshold.checked 14 Nov 2026

Security

The opened repository page does not provide a product security or compliance statement.checked 14 Nov 2026

Setup

The repository instructs users to create its Conda environment and install the package in editable mode with pip.checked 14 Nov 2026

Support

For code questions, the README names Matthias Bitzer and Cen-You Li as contacts.checked 14 Nov 2026