Life is Not Fair, but Your ML Pipeline Can Be

Life is Not Fair, but Your ML Pipeline Can Be

Explore how to evaluate and incorporate trustworthiness and fairness into multi-cloud Kubeflow workflows in this informative conference talk. Gain an overview of responsible AI and fairness, understanding its various aspects and how to assess fairness in both input data and models. Watch demonstrations on using open-source tools like MinDiff, TensorFlow, and ai-widgets for fair model learning, and learn how to include fairness as a parameter in Kubeflow's hyperparameter tuning. Discover the process of transitioning from experimentation in notebooks to production ML pipelines. Obtain valuable recommendations for ensuring trustworthy and fair AI workflows, along with suggestions for Kubeflow enhancements to accelerate this evolution.