Section: New Results

Visual Analytics of EA Data

Participants : Jean-Daniel Fekete, Évelyne Lutton [correspondant] .

Figure 6. Visual Analytics of EA Data: GraphDice showing fitness versus generation view. Red points correspond to the first 10 generations, yellow points, to the 10 last ones, and green points, to the best fitness areas.

An experimental analysis of Evolutionary Algorithms (EAs) usually generates a huge amount of multidimensional data, including numeric and symbolic data. It is difficult to efficiently navigate in such a set of data, for instance to be able to tune the parameters or evaluate the efficiency of some operators. Usual features of existing EA visualisation systems consist in visualising time- or generation-dependent curves (fitness, diversity, or other statistics). When dealing with genomic information, the task becomes even more difficult, as a convenient visualisation strongly depends on the considered fitness landscape. In this latter case the raw data are usually sets of successive populations of points of a complex multidimensional space.

The purpose of this study was to evaluate GraphDice on complex sets of EA data (for artificial and real test-cases), and to sketch future developements of this tool, in order to better adapt it to the needs of EA experimental analysis (Fig. 6 ). An output of this study is the acceptation of the EASEA-Cloud ANR-Emergence project, in which developements will aim at adding tools in GraphDice specific for:

• visualisation of Evolutionary Algorithms written in the EASEA language,

• monitoring the execution of these algorithms on a cloud of computers (CPU + GPU).