Life is not fair, but your ML pipeline can be!
Talkat KubeCon + CloudNativeCon North America 2021 · on Friday, 2021-10-15· at 16:30· 411 Theater + Online·
summary
Shows concrete steps to measure and enforce fairness and trustworthiness inside multi‑cloud Kubeflow pipelines, with a live pipeline demo using MinDiff, TensorFlow and ai‑widgets to detect and mitigate skew in training data and model outputs. Demonstrates adding fairness as a tuning parameter to automated hyperparameter searches in Kubeflow and the code pattern to move experiments from notebooks into reproducible pipeline components. Ends with actionable Kubeflow extension recommendations and integration notes for production.
AI-written summary · about
chapters (7)
- 0:00Introduction to ML fairness
- 4:10Key challenges in ML pipelines
- 10:06Real-world fairness examples
- 14:24Fairness techniques and tools
- 18:04Reducing bias in models
- 22:17Practical fairness strategies
- 25:55Q&A on ML fairness
related talks
30mStand up for ethical AI! How to detect and mitigate AI bias using Kubeflow
05mLightning Talk: Build a Reproducible ML Workflow with Kubeflow Pipelines
39mOperating Deep Learning Pipelines Anywhere Using Kubeflow
18mAutomate the boring ML stuff with pipelines
35mOrganizing data engineering around Kubeflow
23mRemoving Unfair Bias in the Machine Learning Pipeline
37mEnd-to-End ML Pipelines with KubeFlow and TensorFlow Extended (TFX)
54mScaling Machine Learning Pipelines in Cloud