🏂 CARVETM stands for Cluster-Aware Routines for Versatile Embedding and provides a flexible toolbox of implementations of algorithms for tailored and effective application of cluster-aware embedding techniques. The algorithms and code were developed and written by Dr. Amanda Buch and Dr. Logan Grosenick in the Grosenick lab at Weill Cornell Medicine.
CARVE is a novel framework that simultaneously performs joint clustering and embedding by combining standard embedding methods with a convex clustering penalty in a modular way.
⚡⚡⚡
📬 Please join our mailing list to stay up-to-date on new major releases for CARVE. Sign up here!
⚡⚡⚡
🔅 Currently CARVE is composed of the following algorithms:
- Pathwise Clustered Matrix Factorization (PCMF). PCMF implements cluster-aware principal component analysis on a single-view dataset. (Please cite Buch et al., AISTATS 2024 paper)
- Locally Linear Pathwise Clustered Matrix Factorization (LL-PCMF). LL-PCMF implements cluster-aware locally linear embedding on a single-view dataset. (Please cite Buch et al., AISTATS 2024 paper)
- Pathwise Clustered Canonical Correlation Analysis (P3CA). P3CA implements cluster-aware canonical correlation analysis on two-view datasets. (Please cite Buch et al., AISTATS 2024 paper)
💻 Our Github repo implementing algorithms from the AISTATS paper is here: https://github.com/carve-ai/PCMF. Our Main Github repo for CARVE will be linked here following the beta phase.
CARVE is currently proprietary and in beta phase (see license). It will soon be released and licensed for academic use.
All Rights Reserved. Copyright (2022-present) Amanda M. Buch, Conor Liston, & Logan Grosenick
- 📄 Full Paper May 2024: Buch, Amanda M., Conor Liston, and Logan Grosenick. (2024) Simple and Scalable Algorithms for Cluster-Aware Precision Medicine. Proceedings of The 27th International Conference on Artificial Intelligence and Statistics (AISTATS), PMLR 238:136-144 https://proceedings.mlr.press/v238/buch24a/buch24a.pdf
- 📄 Short Paper December 2023: Buch, Amanda M., Liston, Conor & Grosenick, Logan. (2023). Cluster-Aware Algorithms for AI-Enabled Precision Medicine. Neural Information Processing Systems Conference: LatinX in AI (LXAI) Research Workshop 2023, New Orleans, Louisiana. https://doi.org/10.52591/lxai2023121011
- 📄 Preprint November 2022: Buch, Amanda M., Liston, Conor & Grosenick, Logan. Simple and Scalable Algorithms for Cluster-Aware Precision Medicine. AISTATS 2024. https://arxiv.org/abs/2211.16553