YingfanWang / PaCMAP

PaCMAP: Large-scale Dimension Reduction Technique Preserving Both Global and Local Structure
Apache License 2.0
543 stars 54 forks source link

Metric learning with PaCMAP? #52

Open neuro30 opened 1 year ago

neuro30 commented 1 year ago

Would it be possible to add a feature that allows one to embed an unseen/unlabeled data point to an existing embedding using metric learning? This would be similar to the function currently available with UMAP: https://umap-learn.readthedocs.io/en/latest/supervised.html

Thanks!

chipsin87 commented 1 year ago

If this could be implemented, it would be incredible. I absolutely love PaCMAP and have been using the dimensions as predictor variables for supervised models. This would take things to a new level

hyhuang00 commented 1 year ago

Thank you for your feedback! This feature is a top priority on our to-do list, and we are committed to working on it! We will keep you updated on our progress.