Ricerc@Sapienza

The interest on introducing robots in schools has increased significantly in recent years. Robots in these environments are managed by educators who design teaching activities where the students can consolidate the knowledge acquired in the classroom by interacting with the robot. In this context, the use of multiple modalities of communication can become a determining factor to achieve the success of the interaction and a better learning experience.

This paper presents an image fusion method which uses an image dependent multiscale decomposition and a fusion rule which is based on the local image multiscale activity. The latter is used for determining a proper partition of the frequency plane where the max-based fusion strategy is applied; the same image activity is used for guiding the max-based fusion rule. Multiscale local image activity is computed using an estimation of the local Lipschitz regularity at different resolutions.

In this paper local and non-local denoising methods are jointly employed in order to improve the visual quality of the final denoised image. Based on the evidence that the output images of non local denoising methods are not pointwise better everywhere than the outputs images of local methods and than the noisy image itself, the cascade of two improvement steps is applied to the output image of a non local denoising method. The first step aims at correcting the output image by recovering the lost information directly from the noisy one.

The paper presents a method for color quantization (CQ)which uses visual contrast for determining an image-dependent colorpalette. The proposed method selects image regions in a hierarchicalway, according to the visual importance of their colors with respect to thewhole image. The method is automatic, image dependent and requires amoderate computational effort. Preliminary results show that the qual-ity of quantized images, measured in terms of Mean Square Error, ColorLoss and SSIM, is competitive with some existing CQ approaches.

Declarative process mining is the set of techniques aimed at extracting behavioural constraints from event logs. These constraints are inherently of a reactive nature, in that their activation restricts the occurrence of other activities. In this way, they are prone to the principle of ex falso quod libet: they can be satisfied even when not activated. As a consequence, constraints can be mined that are hardly interesting to users or even potentially misleading.

A directed multigraph is said vulnerable if it can generate Braess paradox in traffic networks. In this paper, we give a graph–theoretic characterisation of vulnerable directed multigraphs. Analogous results appeared in the literature only for undirected multigraphs and for a specific family of directed multigraphs. The proof of our characterisation provides the first polynomial time algorithm that checks if a general directed multigraph is vulnerable in O(| V| · | E|2).

A graph G = (V,E) is a pairwise compatibility graph (PCG) if there exists an edge-weighted tree T and two non-negative real numbers dminand dmax, dmin≤ dmax, such that each node u∈V is uniquely associated to a leaf of T and there is an edge (u, v) ∈ E if and only if dmin≤ dT(u, v) ≤ dmax, where dT(u, v) is the sum of the weights of the edges on the unique path PT(u, v) from u to v in T. Understanding which graph classes lie inside and which ones outside the PCG class is an important issue. Despite numerous efforts, a complete characterization of the PCG class is not known yet.

Hand rehabilitation therapy is fundamental for post-stroke or post-surgery impairments. Traditional rehabilitation requires the presence of a therapist for executing and controlling therapy: this implies high costs, stress for the patient, and subjective evaluation of the therapy effectiveness. Alternative approaches, based on mechanical and tracking-based gloves, have been recently proposed.

Peer-assessment entails, for students, a very beneficial learning activity, from a pedagogical point of view. The peer-evaluation can be performed over a variety of peer-produced resources, the principle being that the more articulated such resource is, the better. Here we focus, in particular, on the automated support to grading open answers, via a peer-evaluation-based approach, which is mediated by the (partial) grading work of the teacher, and produces a (partial, as well) automated grading. We propose to support such automated grading by means of a method based on the K-NN technique.