DataLab 2: Fairness metrics, and debiasing techniques for image data

In today's DataLab session, you are going to apply your newly acquired Responsible AI knowledge to the Open Images dataset.

Learning objectives

  1. Propose, and apply appropriate fairness metrics to the Open Images dataset.
  2. Apply a debiasing technique to the Open Images dataset (e.g., fairness through unawareness).

Table of contents:

  1. Q&A and standup: 1 hour
  2. Workshop: 5 hours
  3. Reflection & Work/learning log: 1 hour

Questions or issues?

If you have any questions or issues regarding the course material, please first ask your peers or ask us in the Q&A in Datalab!

Tip: Note down any important questions you might have!

Good luck!

1) Q&A and Standup:

1a Ask questions regarding the independent study material.

1b Answer the following questions:

  • What did you do yesterday?
  • What will you do today?
  • Are there any impediments in your way?

2) Workshop: Open Images

OPen Images

Figure 1. Selection of images, and annotations from the Open Images V6 dataset.

2a Proceed with the Datalab exercises listed in the Jupyter Notebook template.

If you have not yet downloaded the template for the Responsible AI Datalabs, please do so now. You can find it, here.

Note: Need further information or have questions? The lecturers will be available throughout the DataLab session.

3) Reflection & Work/learning log (16:00-17:00):

3a Fill in your work/learning log.

3b Choose, and provide an answer to at least of the following questions:

  1. What surprised you today, and why?
  2. What is the most important thing you learned today? Why do you think so?
  3. What do you want to learn more about, and why?
  4. When were you the most creative, and why do you think that is?
  5. What made you curious today? How does learning feel different when you are curious?
  6. When were you at your best today, and why?
  7. (Assuming we were studying the same thing and you could decide and have access to anything), where would you start tomorrow? Why?
  8. What can/should you do with what you know?

Resources