Easy Fairish: Basics of Classification and Bias


Details
Hi, everybody! Join us for an interesting and relevant workshop on error and bias in classification algorithms.
======================================
Workshop Level: Beginner.
======================================
What you'll learn: We will try easy AI models for classification, which can be executed with a few lines of code. We will show that building models may be too easy: high accuracy may be fast to obtain, but it can conceal high error and bias. We will then review basic methods to assess error and bias. With precoded templates, we will use visualizations of error and bias that remain simple, and accessible to stakeholders with limited AI expertise. Finally, we will reflect on why error and bias should be made transparent and accessible, and be discussed with a diverse team of stakeholders.
======================================
How to join: On the morning of the event, you will get a link via Meetup. Make sure that you have opted for emails from Meetup.
======================================
Speaker: Emma Beauxis-Aussalet
https://www.linkedin.com/in/emma-beauxis/
Emma Beauxis-Aussalet is assistant professor of ethical computing at VU Amsterdam. She has a PhD from Utrecht University, and Masters in Computer Science and Communication. She has worked as a designer, R&D engineer, and project leader in AI for social good. She has researched the means to assess and communicate AI mechanisms, error, and bias. She thrives to enable data-driven systems that are more responsible, ethical, transparent, and made accessible to a large public with or without technical background.
======================================
GitHub Repo: https://bit.ly/31z8J2i
======================================
Agenda: 18:30 - 20:00 Workshop
======================================
If you have any questions, please write to us at amsterdam@pyladies.com.
We are looking forward to meeting you!

Easy Fairish: Basics of Classification and Bias