Skip to content
skwdro Documentation
⌘ K
skwdro Documentation

Bird's eye view

  • Welcome to the docs
  • Getting started
  • Quick tour of WDRO
  • Sinkhorn regularisation of WDRO: SkWDRO
  • User guide

PyTorch part of the library

  • Introduction to the PyTorch interface
  • Some torch examples for practice
    • Polynomial regression
    • Comparison between some classification techniques
    • Simple Neural Network
    • Comparison between some regression techniques
  • Diving into the library and its settings
    • Effect of the epsilon Sinkhorn regularization parameter
    • Understanding the landscape of the lambda optimisation
    • Study the relationship between ERM, WDRO, and SkWDRO
    • Comparison with the python-dro package
  • Illustration on iWildsCam

Scikit part of the library

  • How to use the available scikit-learn estimators
    • skwdro.linear_models.LogisticRegression
    • skwdro.linear_models.LinearRegression
    • skwdro.operations_research.NewsVendor
    • skwdro.operations_research.Portfolio
  • Some visual illustrations for scikit-learn estimators
    • Spatial perturbations and logistic regression
    • Logistic regression
    • Distributionally Robust Portfolio optimization
    • Linear regression
    • Robustify a neural-net on spatial perturbations of the “moons” non-convex dataset

Advanced topics

  • Samplers tutorial
  • Cost functionals tutorial
  • About the tuning of the uncertainty radius
  • More on specific solvers available through scikit-learn estimators
    • skwdro.solvers.entropic_dual_torch.solve_dual_wdro
    • skwdro.solvers.specific_solvers.WDRONewsvendorSpecificSolver
    • skwdro.solvers.specific_solvers.WDROLogisticSpecificSolver
    • skwdro.solvers.specific_solvers.WDROLinRegSpecificSolver
    • skwdro.solvers.specific_solvers.WDROPortfolioSpecificSolver

API

  • skwdro
    • skwdro package
  • Main APIs
    • skwdro.torch module
    • skwdro.base.losses_torch module
    • skwdro.base.samplers.torch module
    • skwdro.base.costs_torch module
    • skwdro.linear_models package
    • skwdro.operations_research package
    • skwdro.base.cost_decoder module
    • skwdro.solvers package
    • skwdro.distributions package
    • skwdro.tests package
skwdro Documentation
/
Main APIs

Main APIs#

API

Here are the main building blocks of skwdro.

Main interface

  • skwdro.torch module
    • robustify()

Build your own robust model

  • skwdro.base.losses_torch module
  • skwdro.base.samplers.torch module
    • References
  • skwdro.base.costs_torch module

Some pre-implemented models

  • skwdro.linear_models package
    • Module contents
  • skwdro.operations_research package
    • Module contents

Miscs/helpers

  • skwdro.base.cost_decoder module
  • skwdro.solvers package
    • Submodules
    • skwdro.solvers.entropic_dual_torch module
    • skwdro.solvers.hybrid_opt module
    • skwdro.solvers.optim_cond module
    • skwdro.solvers.oracle_torch module
    • skwdro.solvers.result module
    • skwdro.solvers.specific_solvers module
    • skwdro.solvers.utils module
    • Module contents
  • skwdro.distributions package
    • Submodules
    • skwdro.distributions.dirac_distribution module
    • Module contents
  • skwdro.tests package
    • Submodules
    • skwdro.tests.test_common module
    • Module contents

© 2024 skwdro team Built with Sphinx 9.0.4