Git Product home page Git Product logo

tfg-ensuring_fairness_in_ml's Introduction

TFG - Tools to Ensure Fairness in Machine Learning

Repositorio de mi trabajo de fin de grado para el Doble Grado en Ingeniería Informática y Matemáticas de la Universidad de Granada sobre un estudio experimental para diferentes modelos de equidad contrafactual. Puede descargar una versión compilada de la memoria en este enlace. Puede descargar las diapositivas de la presentación realizada al tribunal aquí (plantilla cedida por Julio Mulero de la Universidad de Alicante).

Un tutorial para la ejecución del experimento basado en Jupyter Notebook puede ser consultado en el siguiente enlace. Si desea ejecutarlos en su ordenador, será necesario que descargue las dependencias a los siguientes paquetes:

  • Pandas 1.2.4
  • NumPy 1.19.2
  • Scikit-learn 0.24.2
  • matplotlib 3.4.3
  • seaborn 0.11.2
  • pathlib2 2.3.6
  • Aequitas 0.42.0
  • PyStan 2.19.1.1
  • PyMC3 (opcional; para otras_pruebas.py)

Puede hacerlo, bien manualmente, o bien puede utilizar el siguiente comando:

python -m pip install -r requirements.txt

Mejoras en el rendimiento

En un futuro, se propone mejorar los resultados predichos para el modelo de regresión, utilizando un modelo no lineal que se adapte a las distribuciones estudiadas. Por ejemplo, utilizando un modelo de regresión polinomial o segmentada, o redefiniendo el problema en el ámbito de la clasificación y utilizando el algoritmo SVM o una red neuronal para predecir los resultados de las etiquetas.

Description

It proposes as a challenge, after an exhaustive study of the existing bibliography and tools on the subject, to carry out a comparative analysis in Python that includes fairness evaluations of all the families and visualisations for the interpretation of results and models.

To do so, we will start by carrying out an exhaustive review of the different mathematical formalisations of the concept of fairness in the literature, analysing the advantages and disadvantages of each one both from a theoretical point of view and in terms of their practical implementation in specific classification problems by means of automatic learning. In addition, a study and classification of the most popular published algorithms that work with fairness will be carried out and a study of the advantages and disadvantages of each one will be made.


Main Bibliography

  • Barocas, S., Hardt, M., & Narayanan, A. (2018). Fairness and Machine Learning. fairmlbook. org, 2018. URL: http://www.fairmlbook.org.

  • Mitchell, S., Potash, E., Barocas, S., D'Amour, A., & Lum, K. (2018). Prediction-based decisions and fairness: A catalogue of choices, assumptions, and definitions. arXiv:1811.07867.

  • Zafar, M. B., Valera, I., Gomez Rodriguez, M., & Gummadi, K. P. (2017, April). Fairness beyond disparate treatment & disparate impact: Learning classification without disparate mistreatment. In Proceedings of the 26th international conference on world wide web (pp. 1171-1180).

  • M. J. Kusner, J. R. Loftus, C. Russell & R. Silva (2018). Counterfactual Fairness.

  • Saleiro, P., Kuester, B., Hinkson, L., London, J., Stevens, A., Anisfeld, A., ... & Ghani, R. (2018). Aequitas: A bias and fairness audit toolkit. arXiv:1811.05577.

tfg-ensuring_fairness_in_ml's People

Contributors

danibolanos avatar

Watchers

James Cloos avatar  avatar

Forkers

andrefmb

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. 📊📈🎉

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google ❤️ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.