Life is not fair, but your ML pipeline can be!


Authors:   Meenakshi Kaushik, Neelima Mukiri


This session will talk about to evaluate and add trustworthiness and fairness to your multi-cloud Kubeflow workflows. The session will describe the following: * Overview of responsible AI & fairness. * What are the different aspects of fairness. * How to evaluate fairness in the input data & the model. * Demo: Using open source tools such as MinDiff/Tensorflow/ai-widgets for fair model learning * Demo: Including fairness as a parameter in HyperParameter tuning in Kubeflow * Transition from experimentation in Notebooks to ML production pipelines. * Recommendations to ensure your AI workflows are trustworthy and fair. Suggestions for enhancements in Kubeflow that can accelerate this evolution.