Researchers have developed Denoised Conformal Alignment, a novel method for reliably selecting subsets of individuals for treatment based on predicted conditional average treatment effects (CATE). This approach addresses the issue where standard conformal guarantees may not hold for selected subsets. The method constructs proxy errors from pseudo-outcomes and incorporates a denoising step to account for heteroskedasticity, improving power while maintaining false discovery rate control. AI
IMPACT Enhances reliability in selecting individuals for treatment based on AI predictions, potentially improving the effectiveness of interventions.
RANK_REASON The cluster contains a submitted academic paper on a statistical machine learning method.
- alphaXiv
- arXiv
- Benjamini--Hochberg
- CatalyzeX Code Finder for Papers
- Conditional average treatment effect estimation with marginally constrained models
- CORE Recommender
- DagsHub
- Denoised Conformal Alignment
- Gotit.pub
- Hugging Face
- ScienceCast
- Influence Flower
AI-generated summary · Google Gemini · from 2 sources. How we write summaries →
