Strategies for Winnowing Microarray Data

The analysis of microarray datasets is complicated by the magnitude of the available information. Most data mining techniques are significantly hampered by irrelevant or redundant information. Hence it is useful to reduce datasets to manageable size by discarding such useless information. We present techniques for winnowing microarray datasets using singular value decomposition and semidiscrete decomposition, and show how they can be tuned to extract some information about the internal correlative structure of large