manifold-learning methods for dimension reduction

[Shriphani Palakodety]

I implemented MDS and Isomap some time back for core.matrix matrices. I've heard talk of a clojure machine learning framework. Manifold learning approaches complement algorithms like PCA, Kernel PCA and the like and libs like scikit-learn contain good implementations of these algorithms.

https://github.com/shriphani/clojure-manifold

[Mike Anderson]

Looks great, thanks for sharing!

Do you have any demo code that shows an example of usage? This would be useful for people to get a feel of the code, and also to check how the interface might fit within a broader Clojure machine learning framework.

[Shriphani Palakodety]

There's an explanation of the algorithms here: http://blog.shriphani.com/2014/10/29/low-dimension-embeddings-for-visualization/ and here: http://blog.shriphani.com/2014/11/12/the-isomap-algorithm/

I have added more info to the repo README: https://github.com/shriphani/clojure-manifold/blob/master/README.md

Scikit-learn typically contains a lot of useful info about the models as well but this repo is a bit haphazard. I'll clean the docs up in a bit.