SHAP beyond the standard graphics: co-design of ML-models in earth sciences
summary
Covers using SHAP feature-attribution values combined with dimensionality reduction and clustering to identify drivers of ML model predictions in earth sciences, and how those techniques can be used to involve stakeholders in co-designing models. Describes practical case studies applying this pipeline and lessons learned for leveraging SHAP beyond standard visualizations to produce actionable information for non-technical participants.
AI-written summary · about
related talks
31mBeyond the Black Box: Interpreting ML Models with SHAP
31mWhat We Expect from XAI - A scientist’s experience
42mMitigating Bias in Models with SHAP and Fairlearn
27mInterpretability and explainability in machine learning
35mSHAPtivating Insights: unravelling blackbox AI models
33mBringing life and motion to AI Explainability
35mTaking a peek under the hood - Interpreting black box model
01h 36mOpen the Black Box: an Introduction to Model Interpretability with LIME and SHAP