Sign in to confirm you’re not a bot
This helps protect our community. Learn more
These chapters are auto-generated

Introduction

0:00

Agenda

1:00

Responsible AI

3:20

Why is responsible AI important

5:55

Debugging AI models

11:25

Data representation

16:57

Data analysis

18:31

Classification use case

20:35

Data transformation

22:20

How to use the responsible AI dashboard

23:17

Feature metadata container

24:30

Using the REI instant instance

25:07

Defining the use case

25:48

Cohorts

28:31

Decision Tree

31:23

Heatmap

33:10

Highlight

34:13

Error Rate

34:58

Error Coverage

35:55

Error Rates

40:15

Comparing Cohorts

41:31

Probability Distribution Social Graph

45:05

Metrics

46:25

Data

51:54
How to expose ML model errors, data bias & interpretability with responsible AI
The astonishing growth in AI innovations are transforming our lives and society. Companies are adopting AI in their business process and products to gain a competitive advancement. The rapid advancements are seen across many industries such as finance, healthcare, education, manufacturing, etc. As society expectation for AI are evolving, there's increasing scrutiny on what harms AI systems can cause with no transparency or accountability enforced. As a result, there’s growing government compliance regulations on AI in some industries. On the other hand, data scientists, AI developers and decision-makers face the challenge of finding the right tools to enable them to analyze machine learning models for fairness, safety & reliability, explainability and accountability. In this session, we will explore the Azure Machine Learning's Responsible AI dashboard tool that enables data scientists and companies to analyze and debug AI systems to be less harmful to society, more trustworthy and meet compliance requirements. Useful Resources: • Diabetes hospital readmission (Demo): https://github.com/ruyakubu/responsib... • Responsible AI Toolkit (OSS): https://github.com/microsoft/responsi... • E2E Blog tutorial: https://aka.ms/TCBlogRAIGettingStarted • Responsible AI information: https://aka.ms/rai-dashboard [eventID:20273]

Follow along using the transcript.

Microsoft Reactor

114K subscribers
Live chat replay is not available for this video.