{"slug": "metric-aware-pca-as-a-linear-instance-of-geometric-deep-learning", "title": "Metric-Aware PCA as a Linear Instance of Geometric Deep Learning", "summary": "Metric-Aware PCA (MAPCA) parameterizes principal component analysis by a positive-definite metric matrix, with a canonical subfamily interpolating between standard PCA and output whitening. The paper positions MAPCA within the geometric deep learning framework, establishing that the metric serves as a geometric prior, the orthogonal group preserving it as the induced symmetry group, and MAPCA solutions as equivariant under this group. A uniqueness theorem characterizes Invariant PCA as the unique linear data-derived metric in the MAPCA family that is equivariant under arbitrary diagonal rescaling, with the paper extending the framework to kernel PCA, spectral graph methods, and deep MAPCA constructions.", "body_md": "arXiv:2605.27456v1 Announce Type: new\nAbstract: Geometric deep learning organises neural architectures around the symmetries of their data domain, with the choice of symmetry group serving as a geometric prior that determines what representations can be learned. Metric-Aware Principal Component Analysis (MAPCA) parameterises principal component analysis by a positive-definite metric matrix, with a canonical subfamily interpolating between standard PCA and output whitening and a diagonal-metric point recovering Invariant PCA (IPCA). This paper positions MAPCA within the geometric deep learning framework. The metric is read as the geometric prior; the orthogonal group preserving it is the symmetry group it induces; MAPCA solutions are equivariant under this group with the resulting spectrum invariant; and MAPCA's defining constraint is the linear analogue of the Schur-type weight constraints used in equivariant networks. Across six axes - domain, symmetry group, equivariance, invariance, architectural primitive, and geometric prior - we construct a precise dictionary between MAPCA and geometric deep learning. The technical anchor is a uniqueness theorem characterising IPCA as the unique linear data-derived metric in the MAPCA family that is equivariant under arbitrary diagonal rescaling and projects onto the fixed-point set of the action, equivalent under normalisation to the variance-maximisation criterion in its precise form. The paper closes with three bridges: kernel PCA as the nonlinear extension, spectral graph methods as MAPCA on graphs, and a deep MAPCA construction extending the positioning into deep equivariant networks", "url": "https://wpnews.pro/news/metric-aware-pca-as-a-linear-instance-of-geometric-deep-learning", "canonical_source": "https://arxiv.org/abs/2605.27456", "published_at": "2026-05-28 04:00:00+00:00", "updated_at": "2026-05-28 04:28:46.210949+00:00", "lang": "en", "topics": ["machine-learning", "neural-networks", "ai-research"], "entities": ["Principal Component Analysis", "Metric-Aware Principal Component Analysis", "Invariant PCA", "Geometric Deep Learning"], "alternates": {"html": "https://wpnews.pro/news/metric-aware-pca-as-a-linear-instance-of-geometric-deep-learning", "markdown": "https://wpnews.pro/news/metric-aware-pca-as-a-linear-instance-of-geometric-deep-learning.md", "text": "https://wpnews.pro/news/metric-aware-pca-as-a-linear-instance-of-geometric-deep-learning.txt", "jsonld": "https://wpnews.pro/news/metric-aware-pca-as-a-linear-instance-of-geometric-deep-learning.jsonld"}}