Addressing Instrument-Outcome Confounding in Mendelian Randomization through Representation Learning

ORID zxJXgfCm63 · tags icml2026-repro paper-zxJXgfCm63

#StatusPageArtifactClaim excerpt
1VERIFIED 2/201-autoencoder-satisfying-reconstruction-cross-enviartifactTheorem 3.1 shows that an autoencoder satisfying reconstruction and cross-enviro…
2VERIFIED 2/202-extends-identification-general-diffeomorphic-smoartifactTheorem 3.2 extends identification to general diffeomorphic (smooth, invertible)…
3VERIFIED 2/203-establishes-valid-instrument-bijective-measurablartifactProposition 2.2 establishes that if Z is a valid instrument and κ is a bijective…
4VERIFIED 2/204-semi-synthetic-experiments-use-genetic-variantsartifactSemi-synthetic experiments use 652 genetic variants from the GLP1R region across…
5VERIFIED 2/205-proposed-sls-sls-estimators-achieveartifactThe proposed 2SLS(Ŵ) and PO(V̂)-2SLS(Ŵ) estimators achieve near-zero bias across…
6VERIFIED 2/206-partial-identification-remains-unbiased-whileartifactPartial identification (p̂ p) remains unbiased, while over-specifying the late…

Open logbook index · logbook.json