Duration: 20:10

1,116 views

0

Speakers:
Besmira Nushi, Principal Researcher, Microsoft Research
Mehrnoosh Sameki, Senior Program Manager, Microsoft Azure Machine Learning
Amit Sharma, Senior Researcher, Microsoft Research India

Assessing and investigating machine learning (ML) models prior to deployment remains at the core of developing trustworthy and responsible artificial intelligence (AI). While different open-source tools have been proposed for assessing fairness, explainability, or errors of an ML model, these properties are not independent, and ML practitioners may need several of these functionalities to fully identify, diagnose, mitigate issues, and take action in the real world. In this session, we will demonstrate the Responsible AI Toolbox. This toolbox was built with two intentions: accelerate the development lifecycle for ML in a way that implements and applies Responsible AI principles, and serve as a collaboration framework for research in the Responsible AI field. We will introduce the overall workflow, from the ease of configuring the interoperable dashboards up to the intended experience. We will showcase how the toolbox can be used to assess models through a responsible AI lens and to analyze data for causal decision-making with the goal of identifying actions that can impact desirable outcomes in the real world. Attendees will be able to access the different parts of the demo through online interactive deployments of the toolbox on illustrational datasets and models.

Resources: https://github.com/microsoft/responsible-ai-widgets/

Learn more about the 2021 Microsoft Research Summit: https://Aka.ms/researchsummit