[2604.00485] The Rashomon Effect for Visualizing High-Dimensional Data

Computer Science > Machine Learning arXiv:2604.00485 (cs) [Submitted on 1 Apr 2026] Title:The Rashomon Effect for Visualizing High-Dimensional Data Authors:Yiyang Sun, Haiyang Huang, Gaurav Rajesh Parikh, Cynthia Rudin View a PDF of the paper titled The Rashomon Effect for Visualizing High-Dimensional Data, by Yiyang Sun and 3 other authors View PDF HTML (experimental) Abstract:Dimension reduction (DR) is inherently non-unique: multiple embeddings can preserve the structure of high-dimensional data equally well while differing in layout or geometry. In this paper, we formally define the Rashomon set for DR -- the collection of `good' embedding -- and show how embracing this multiplicity leads to more powerful and trustworthy representations. Specifically, we pursue three goals. First, we introduce PCA-informed alignment to steer embeddings toward principal components, making axes interpretable without distorting local neighborhoods. Second, we design concept-alignment regularization that aligns an embedding dimension with external knowledge, such as class labels or user-defined concepts. Third, we propose a method to extract common knowledge across the Rashomon set by identifying trustworthy and persistent nearest-neighbor relationships, which we use to construct refined embeddings with improved local structure while preserving global relationships. By moving beyond a single embedding and leveraging the Rashomon set, we provide a flexible framework for building interpretab...