clustering using mutual information as metric

Question

I have a set of data that I would like to cluster.

Considered some domain features, I think that Mutual Information is a pretty good measure of how much to elements of the dataset are close one to the other.

Each data point in my dataset is a vector of categorical features.

Have anyone experience with such approach?

Answer 1

There are many clustdeing algorithms that can be used sith any similarity or dissimilarity. E.g.,

1. Hierarchical Agglomerative Clustering
2. DBSCAN
3. OPTICS

They are commonly used with distances, but they don't require any special properties, so they can be made to work with similarities, too.