Abstract
The work described in this paper covers mainly the exploration of an important paradigm called amorphous computing. With the current smart systems composed of a great number of cognitive entities, amorphous computing offers useful tools and languages to emerge a coherent behavior relying on local communications and limited capabilities. In order to emphasize its capabilities, the problem of clustering microarray data has been solved within this new computing paradigm. Moreover, it is difficult and time consuming to deal with a large amount of noisy gene expression data. The core motivations of amorphous computing come from self-assembly property to emerge clusters of complex gene expressions. In particular, the process of clustering was applied with respect to the Growing Neural Gas algorithm (GNG), which is an incremental learning method and a visual technique. Although the GNG draws important features from the Self-Organizing map (SOM), it yields accurate results when no information about the initial distribution is available. This contribution considers a huge number of amorphous computing entities placed irregularly, each with a randomly selected reference vector. Using the GNG, the visualization of clusters of gene expressions is obtained by amorphous computing particles. The results obtained using the Netlogo platform are very encouraging and argue that self-organization by means of local interactions and irregularly placed particles will be qualified with performance in a real amorphous computing system.