Boolean Matrix Factorizations and Data Mining

matrix as a Boolean product of two (possibly smaller)
binary matrices. It is a powerful tool that has found
applications in varied fields like psychometrics,
extremal combinatorics, communication complexity,
and logic circuit design. In this talk we will
approach Boolean matrix factorizations from data
miner's perspective. In data mining, Boolean matrix
factorizations can be seen as a logical continuation
of a line of research starting from the frequent itemset
mining and extending to the modern pattern set mining
algorithms. Unlike many of the other fields, in data
mining the interest is chiefly on approximate
factorizations for better handling of noisy data.

In this talk I will discuss about some algorithms for
finding the approximate Boolean matrix factorizations.
Before that, however, I'm going to talk about the
computational complexity of approximate and exact
Boolean matrix factorizations and some related problems.
After the basic algorithms, I will discuss on some extensions,
for example, how to apply the Minimum description length (MDL)
principle to choose the rank of the decomposition and what
can we say about the factorizations of sparse matrices.