Abstract
We propose a novel methodology to extract gene expression profiles from microarray experiments based on the application of an AI process (multiobjective optimization) to establish relationships between each profile and the most suitable existing technique to discover it. We first determine a space of potential hypothesis (techniques) aggregating well-known and widely used methods. Then, for each profile, we perform a multiobjective search over the space of hypothesis in order to find the technique (or aggregation of them) that better finds it.