High throughput measurements, such as microarray based gene expression profiling, offers a more complete view of the biochemical changes associated with cancer. However, more data means more noise, more uncertainty and an explosion of the hypothesis space, all impeding association based learning often applied both in basic and clinical cancer research. Our group is interested in the meaningful and responsible application of high throughput measurements for cancer research. We implemented several methods that increased the reliability of microarray measurements. We are also interested in approaches that combine high throughput measurements in a manner that describe essential biology in a robust fashion, such as developing a gene expression signature of chromosomal instability.