Efficient and Generalizable Archetypal Analysis for Discrete Data
Archetypal Analysis (AA) represents observations as convex combinations of extremal data-driven profiles, yielding interpretable low-dimensional descriptions of complex datasets.
ProofPaper ↗
Key points
- We introduce an efficient likelihood-based framework for AA supporting Bernoulli, Poisson, and multinomial observation models.
- We further introduce a cross-validated predictive likelihood criterion for selecting the number of archetypes, providing a principled alternative to reconstruction-error heuristics and stability-based diagnostics.
- Applications to single-cell RNA sequencing, microbiome composition, and somatic mutation data show that the learned archetypes capture interpretable domain-specific structures while achieving competitive likelihood fits and stable solutions.
- Overall, the proposed framework enables efficient likelihood-based archetypal analysis of discrete data, complemented by predictive likelihood-based model selection.
Sources (1)
- [1]Efficient and Generalizable Archetypal Analysis for Discrete DataarXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 8, 02:27 PM
Archetypal Analysis (AA) represents observations as convex combinations of extremal data-driven profiles, yielding interpretable low-dimensional descriptions of complex datasets.
We introduce an efficient likelihood-based framework for AA supporting Bernoulli, Poisson, and multinomial observation models.
Extractive summary: sentences quoted from the sources.