Reproduce “Connecting Score Matching, Maximum Likelihood, and Expectation-Maximization in Mixed Linear Regression” 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
Zhankun Luo, Abolfazl Hashemi
The paper studies score matching for a two-component mixed linear regression model, where each response comes from one of two covariate-dependent lines. It shows that integrating the objective across a diffusion path is asymptotically equivalent to maximum likelihood, yielding sign-aligned consistency and Gaussian limits. At n=4000, coverage was 95.0%; fixed-scale imbalance signal can vanish with separation for fixed candidates.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.