Researchers have developed a novel geometry-aware KDE model for multiclass quantification, utilizing Aitchison geometry and log-ratio representations. This approach addresses limitations of existing methods that ignore the geometric properties of the probability simplex. The proposed method incorporates shrinkage regularization for improved robustness and offers both point-estimation and Bayesian inference procedures for class prevalences. Experiments across various domains demonstrate its competitiveness with state-of-the-art quantifiers and improvement over standard KDE-based baselines. AI
IMPACT Introduces a novel method for class prevalence estimation that improves upon existing techniques by considering geometric properties.
RANK_REASON The cluster contains an academic paper published on arXiv detailing a new method in machine learning.
- Aitchison geometry
- Alejandro Moreo PhD
- alphaXiv
- arXiv
- CatalyzeX Code Finder for Papers
- Connected Papers
- DagsHub
- Geometry-Aware Bayesian Quantification via Compositional Data Analysis
- Gotit.pub
- Hugging Face
- IArxiv
- Influence Flower
- KDE
- Litmaps
- machine learning
- ScienceCast
- scite Smart Citations
- arXivLabs
AI-generated summary · Google Gemini · from 2 sources. How we write summaries →
