3.1.4 How to measure classification bias to answer fairness

Course subject(s) Module 3: AI in Practice: Preparing for AI

In this video lesson Emma Beauxis-Aussalet, Assistant Professor of Ethical Computing at Vrije Universiteit Amsterdam (VU Amsterdam), will discuss AI errors, and why it is essential to measure them to make compliance possible.

This video lesson is based on scientific studies and talks that are referred to below. Note: This additional information is not mandatory for the course and is primarily intended for learners who wish to dive deeper into the material.

Beauxis-Aussalet & Hardman (2017). Extended Methods to Handle Classification Bias. Int. Conf. Data Science and Advanced Analytics (DSAA).
Beauxis-Aussalet, van Doorn, Hardman (2019). Supporting End-User Understanding of Classification Errors: Visualization and Usability Issues. Journal of Information Science (JoIS).

Talks:

Making Classification Error & Bias Transparent & Understandable (2019).
Old math and simple bar charts to make classification bias understandable (2020), (starts at 22:40).

AI in Practice: Preparing for AI by TU Delft OpenCourseWare is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.
Based on a work at https://online-learning.tudelft.nl/courses/ai-in-practice-preparing-for-ai/ /