Exploring Fairness in Machine Learning for International Development
| Course Details | ||||||
|---|---|---|---|---|---|---|
|
||||||
| Outcomes | ||||||
|
||||||
| Price & Terms | ||||||
|
||||||
| Technical | ||||||
|
||||||
| Specification | ||||||
|
||||||
Short video modules on bias and fairness in machine learning for development, with fairness criteria and case studies on real data.
Machine learning tools used in international development can carry bias into decisions that affect people’s lives. This short set of materials was built to help students and faculty understand bias and fairness in machine learning and its appropriate use, and is designed to fit into an existing course over one to two weeks. Four video modules move through the ethical questions, development examples, a framework for responding to them, and data-based case studies.
What You Will Achieve
- Recognize ethical challenges in machine learning for development
- Apply the USAID appropriate use framework to a project
- Compare fairness criteria and the limits of fairness through unawareness
- Examine gender and demographic bias in data-based case studies
Program Overview
Ethics in Machine Learning
Why bias and appropriate use matter.
- Introduction to ethics in machine learning
- USAID appropriate use framework
- A solar lighting example
Pedagogical Framework
Tools for addressing ethical challenges.
- Protected attributes and fairness through unawareness
- Fairness criteria
Case Studies with Data
Bias in practice.
- Mitigating gender bias
- A case study on pulmonary health
- Demographic bias in NLP
Course Completion
Job Outlook
These ideas matter to data scientists, machine learning engineers and practitioners working on development and social impact projects.
Good for: Instructors and students who want a compact module on fairness in machine learning to add to an existing course.
Keep in mind: Spring 2020 material made of lecture videos in four short modules; it is a teaching add-on of one to two weeks, not a full course in machine learning.
Bottom line: A focused, practical introduction to fairness in machine learning, grounded in development examples.
Frequently Asked Questions
How long does it take?
The four modules are designed to be added to an existing course over a period of one to two weeks.
What case studies are included?
Three, working with data: mitigating gender bias, a case study on pulmonary health, and demographic bias in natural language processing.
Who developed it?
The CITE team created it to build the capacity of students and faculty on bias, fairness and the appropriate use of machine learning.
User Reviews
Write your review
You must be logged in to post a review.
Reviews can only be posted by registered learners. Log in or create a free account, then you will return here to post.
There are no reviews yet.