Random forest weighted local manifold estimation
-
- 0 Rating
- 0 Reviews
- 1 Students Enrolled
Random forest weighted local manifold estimation
A statistical framework using random forest weights to improve manifold estimation on high-dimensional point cloud data.
-
- 0 Rating
- 0 Reviews
- 1 Students Enrolled
Requirements
- Manifold Learning, Random Forest, Dimension Reduction, Non-Euclidean Data, Consistency, Geometry
General Overview
Description
This paper proposes a random forest weighted local manifold estimator designed for high-dimensional point cloud data. It integrates local PCA projections and forest-based adaptive weighting to overcome challenges like curvature variation, noise, and curse of dimensionality. The method constructs data-driven neighborhoods via tree-based splits that maximize local heterogeneity. Theoretical contributions include proofs of pointwise consistency, convergence rates, and asymptotic normality. The approach is motivated by applications in 3D geometry reconstruction, with implications for image processing, 3D modeling, and statistical shape analysis.
Recommended for you
Meet the instructors !
David Siang-Li Jheng is a PhD candidate at the Doctoral School of Cybernetics and Economic Statistics, Bucharest University of Economic Studies, Romania. His research focuses on detecting anomalies and modeling dependence structures in high-dimensional, high-frequency financial data.
With a background in financial engineering and mathematics from National Yang Ming Chiao Tung University (NYCU) and National Taiwan Normal University (NTNU), he investigates systemic risks through advanced methodologies such as Financial Risk Meters and anomaly detection models.