From UWMSR

Post date: Oct 18, 2013 6:31:41 PM

The progrm can be found here: http://www.cs.washington.edu/mld/2013/program.html

Rich Caruana (MSR), "Clustering: Probably Approximately Useless?"

- Meta clustering: using clustering diversity to find the optimal clustering at the end

Marina - Preserving geometry in manifold

Deep learning -- MAP s not necessarily the hypothesis selection.

-- predicate depth and median

BIC for singular models

-- Laplace approximation

Learnign fromcounts:

historical count

- good for MapReduce

- state of the art

Taskar

New summarization n

Multi-Relational Latent Semantic Analysis

-