Ricerc@Sapienza

Neural networks require a careful design in order to perform properly on a given task. In particular, selecting a good activation function (possibly in a data-dependent fashion) is a crucial step, which remains an open problem in the research community. Despite a large amount of investigations, most current implementations simply select one fixed function from a small set of candidates, which is not adapted during training, and is shared among all neurons throughout the different layers. However, neither two of these assumptions can be supposed optimal in practice.

Forecasting electricity prices is today an essential tool in the day-ahead competitive market. Prediction techniques based on neural and fuzzy neural networks are very promising in terms of prediction performance and model accuracy. In this paper, we investigate the applicability to the electricity market of three well-known approaches, namely Radial Basis Function neural networks, Mixture of Gaussian neural networks and Higher-Order Neuro-Fuzzy Inference System.

This paper investigates dynamic variation in the morphologic distribution of dense non-aqueous phase liquids (DNAPLs), which take into account the coupled mass transfer. Experiments were carried out in a 2D tank representing a reconstructed aquifer model. DNAPL dissolution rates were investigated over a wide range of DNAPL saturations, several source configurations, and different hydraulic conditions. Morphometric indexes are presented that take into consideration further factors affecting the dissolution process.