PyePAL: Pareto active learning for Python

This package implements an active learning approach to efficiently and confidently identify the Pareto front with any regression model that can output a mean and a standard deviation.

It works with any number of objectives, missing data, and is highly customizable.

If you find this code useful for your work, please cite:

• Jablonka, K. M.; Giriprasad, M. J.; Wang, S.; Smit, B.; Yoo, B. Bias Free Multiobjective Active Learning for Materials Design and Discovery, ChemRxiv 2020 (10.26434/chemrxiv.13200197.v1).

• Zuluaga, M.; Krause, A.; Püschel, M. E-PAL: An Active Learning Approach to the Multi-Objective Optimization Problem. Journal of Machine Learning Research 2016, 17 (104), 1–32.

Contents

• Getting Started
  • Installation
  • Which class do I use?
  • Running an active learning experiment
    • Hyperparameter optimization
    • Reclassification schedule
    • Logging
    • Properties of the PAL object
    • Exploring a space where all objectives are known
    • Batch sampling
    • Adding new points to the design matrix
    • Plotting a learning curve
  • Caveats and tricks with Gaussian processes
• Background
  • The intuition behind the algorithm
  • How do the hyperparameters influence the algorithm?
• Tutorials
  • 1. One active learning step using GPR models built with GPy
    • Topic covered
  • 2. Active learning with “measure” function and sklearn models
    • Topic covered
  • 3. Quantile regression
  • 4. Plotting a learning curve
• The PyePAL API reference
  • The PAL package
    • Core functions
    • Base class
    • For GPy models
    • For coregionalized GPy models
    • For sklearn GPR models
    • For quantile regression with LightGBM
    • For GPR with GPFlow
    • Schedules for hyperparameter optimization
    • Utilities for multiobjective optimization
    • Utilities for plotting
    • Input validation
  • The models package
    • Helper functions for GPR with GPy
• Developer notes
  • Contribution Guidelines
    • Commit messages
    • Python code
    • New features
    • Releases
  • Implementing a new PAL class

Indices and tables

• Index

• Module Index

• Search Page