I have seen the terms "frequent pattern mining", "subspace clustering", and "biclustering". They all pertain to finding clusters using subsets of the data attributes. What's the difference?
Much of the difference is the communities that use them.
Biclustering comes usually from biology, gene expression levels and such. The patterns they look for have typical characteristics, and they are usually mostly interested in trends (co-expression), not the actual numerical values.
To pattern mining, everything is a pattern, so of course they can do biclustering, too. All you need to solve is the complexity and the definition of what is an interesting pattern... They're special... the issue here is, they may easily miss patterns that are too frequent since these may trigger a "triviality" threshold. And other, more rare patterns, may in turn be considered random and dropped. It's hard to strike the right balance of not discarding the patterns you are interested in, and not producing way too many patterns to be usable (since at the end, everything is a pattern ...).
Subspace clustering is the data mining view. It should come along with thoughts about how to implement this efficiently in a large data base context. They often aren't as sophisticated as the pattern mining methods, but should be so fast that you can just give them a try, and see if they find anything useful. Maybe they do, maybe the don't - there is always a chance that there is nothing to find.