New Semi-Supervised Meta Additive Model Proposed for Robust Machine Learning

A research paper titled "S2MAM: Semi-supervised Meta Additive Model for Robust Estimation and Variable Selection" was published on arXiv with identifier arXiv:2604.19072v1. The work introduces a novel semi-supervised learning framework designed to overcome limitations in traditional manifold regularization approaches. Conventional methods using Laplace-Beltrami operator-based regularization often rely on graph Laplacian matrices that are highly dependent on predefined similarity metrics. This dependency can result in inappropriate penalties when handling redundant or noisy input variables. The proposed S2MAM model employs a bilevel optimization scheme that simultaneously updates similarity matrices while automatically identifying informative variables. The approach maintains interpretability while addressing the geometric structure requirements of Riemannian manifolds in semi-supervised learning contexts. The methodology represents an advancement in machine learning techniques for situations where both labeled and unlabeled data are available. The research contributes to more robust estimation and variable selection processes in data analysis applications.