This paper introduces Wasserstein Tangential PCA (WT-PCA), a novel method for learning principal variations of probability measures within Wasserstein geometry. The approach utilizes a dynamical formulation to interpret log-PCA as a variational method, enabling the capture of geodesic variation modes. The research also derives statistical convergence rates for empirical WT-PCA estimates, particularly in relation to the 2-Wasserstein distance. AI
IMPACT Introduces a new method for analyzing probability distributions, potentially impacting downstream AI tasks that rely on statistical modeling of data.
RANK_REASON The cluster contains an academic paper detailing a new statistical method (WT-PCA) for analyzing probability measures, submitted to arXiv.
- 2-Wasserstein distance
- alphaXiv
- arXiv
- CatalyzeX
- DagsHub
- Gotit.pub
- Hugging Face
- Log-PCA
- optimal transport
- ScienceCast
- Wasserstein geometry
- Wasserstein space
- Wasserstein Tangential PCA (WT-PCA)
AI-generated summary · Google Gemini · from 2 sources. How we write summaries →
