Reducing Unfair Bias in Machine Learning (W7109G-SPVC)

Overview

The need for trust in AI has been of importance and one way of achieving it is through mitigating discrimination and bias in machine learning models throughout the AI application lifecycle. This course will give you an overview on the concept of fairness which helps in building trust in AI and how "AI Fairness 360" open source toolkit can help you implement debiasing techniques to measure, understand and mitigate AI bias. Learners will be provided an overview of AI fairness and bias concepts, how to measure bias in models and how to apply fairness algorithms to reduce unwanted bias. It will also walk you through a demo of working of  "AI Fairness 360" open source tool kit and using this tool kit on a real-world use-case.

 

IBM Clients and Sellers - Consider this course as part of an individual or enterprise subscription service:

  • IBM Data/AI Individual Subscription (SUBR003G)
  • IBM Learning for Data and AI Enterprise Subscription (SUBR004G)
  • IBM Learning Individual Subscription with Red Hat Learning Services (SUBR013G)
  • IBM Learning for Data and AI Individual Subscription (SUBR022G)
  • IBM Learning Individual Subscription with Red Hat Learning Services (SUBR023G)

Audience

This course is intended primarily for Analytics Leaders, Data Science Leaders and Practicing Data Scientists, Machine Learning Engineers and AI specialists. Anyone with an interest in AI Trust and bias mitigation concepts having the prerequisite knowledge required.

Prerequisites

In order to be successful, you should have a basic understanding of data science, machine learning, and Python.

Objective

• Recognize the need of Trustworthy AI 
• Describe and differentiate various factors that can build trust in AI 
• Appraise situations that require a focus on fairness 
• Analyze where unwanted bias comes from 
• Recognize methods to mitigate unwanted bias

Mostra dettagli

Course Outline

The Big Picture of Trustworthy AI and Algorithmic Fairness 
• Recognize the need of Trustworthy AI 
• Describe and differentiate various factors that can build trust in AI 
• Appraise situations that require a focus on fairness 
• Analyze where unwanted bias comes from 
• Recognize methods to mitigate unwanted bias 

Mitigating Bias using AI Fairness 360 
• Describe various bias mitigating algorithms that can be intervened in a ML Pipeline 
• Explain the usage of various bias mitigating algorithms 
• Recognize the role of open source toolkit AI Fairness 360 in mitigating bias 
• Describe various features and capabilities of opensource AI Fairness toolkit 
• Analyze the working of opensource AI Fairness toolkit with an interactive demo 

Use AI Fairness 360 in a Real World Use Case 
• Observe the effects of unfair bias for a real world healthcare use case 
• Apply bias mitigation algorithms to reduce bias in the healthcare use case using the AI Fairness 360 toolkit