| 摘要 |
A self-organizing algorithm for the massive analysis of gene expression data from DNA array experiments, beginning from a structure composed of two "daughter" neurons connected to a "mother" neuron, which are composed of a list of values or profile of initially random data. The algorithm divides the data set into data sub-sets in a successive series of cycles. Each of the cycles has a series of stages in which the data is introduced to the terminal neurons and the latter are updated. When the updating produced between two consecutive stages is below a certain pre-established minimum level, it is considered that the network has converged in that cycle and it is decided it the network should continue to grow or if it has already stopped growing, in which case the algorithm process is finalized. <IMAGE>
|
