Know more

In WP 2, the tool for recognition of emotional states will be created. To achieve this, the responsible PostDoc (PostDoc 2) will start early on during the collection of the database (M4, i.e. in parallel with WP 1), first gathering a list of characteristics of the recognition tool deemed necessary and/or desirable by the partners and associates. This list will refer to both performance (sensitivity, specificity, validity) and software interface-design (required features, ease-of-use). Then, while completing the database, the first analyses will aim at identifying the optimal set of parameters to achieve the desired performance. We will test all combinations of these different parameter sets with a panel of classification methods (cluster-analyses, discriminant function analyses, different artificial neuronal networks (Kohonen, perception of different dimensions and different numbers of layers)), in order to identify the most powerful one. The tool will be perpetually improved as the additional data come in from WP 1, until we will finally have a version ready for validation in WP 3.