Clustering algorithms: on learning, validation, performance, and applications to genomics.

摘要: The development of microarray technology has enabled scientists to measure the expression thousands genes simultaneously, resulting in a surge interest several disciplines throughout biology and medicine. While data clustering been used for decades image processing pattern recognition, recent years it joined this wave activity as popular technique analyze microarrays. To illustrate its application genomics, applied from set groups together those whose levels exhibit similar behavior samples, when samples offers potential discriminate pathologies based on their differential patterns gene expression. Although now many context microarrays, remained highly problematic. choice algorithm validation index is not trivial one, more so applying them high throughput biological or medical data. Factors consider choosing an include nature application, characteristics objects be analyzed, expected number shape clusters, complexity problem versus computational power available. In some cases very simple may appropriate tackle problem, but situations require complex powerful better suited job at hand. paper, we will cover theoretical aspects clustering, including error learning, followed by overview algorithms classical indices. We also discuss relative performance these indices conclude with examples biology.