動画検索
関連広告
検索結果
Introduction
Tip1: Incorporate domain knowledge
Tip 2: Build intuitive features
Tip 3: Reduce correlated features
Tip 4: start with a simple model
Tip 5: Understand your audience
Tip 6: Avoid conclusions about causality
Tip 7: Be aware of confirmation bias
Tip 8: Use multiple methods
Tip 9: Use TreeShap
Tip 10: Python package dependencies
About me
XAI in regulation: "rights to explanation"
Performance-Explainability trade-off
XAI post-hoc algorithm example: LIME
Who are the prototypical users of XAI?
Where we started: Research into XAI Design Practices
Methodology
Challenges for practitioners: in the dark design process
XAI can lead to inequalities of experience
Conclusions: Bridging from algorithms to UX