Reproduce “Causal Representation Learning with Instantaneous and Lagged Relations via Nonstationarity” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Tatsuya Yamada, Hiroshi Morioka, Yoshinobu Kawahara
iCReN uses changes in transition-noise distributions linked to an observed auxiliary variable to learn hidden components and estimate instantaneous and lagged causal relations. It assumes invariant transition and observation functions, plus sufficient variation. In a five-factor synthetic test, iCReN-T’s component-recovery score was 0.951, versus IDOL’s 0.468; linear transitions with instantaneous edges remain a stated limitation.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.