Recent advancements in diffusion-based image generation and large vision/language models have revolutionized visual brain decoding (VBD), driving progress in neuroscience and brain-computer interfaces. While state-of-the-art models produce high-quality reconstructed images, a significant semantic gap remains between original stimuli and reconstructed images, posing challenges for applications like forensics, medical treatments, and human-robot interactions. This gap arises from VBD models’ limitations in interpreting brain signals and generating accurate representations. To address this, we analyze the issue through the lens of salient object detection, statistically comparing the similarity between visual stimuli and reconstructed images with a focus on salient objects. To our knowledge, this is the first study to evaluate fMRI-based VBD models from this perspective. Our findings provide measurable insights to guide the development of VBD models that align more closely with human perception.

Despite progress in visual reconstruction from fMRI signals, evaluating reconstruction quality remains challenging due to noisy, low-resolution data and semantic ambiguity. Conventional metrics often overlook perceptual and structural alignment with the original stimuli. To address this, we propose Graph-based Semantic and Structural Similarity (GSS), a novel evaluation approach that represents both stimuli and reconstructions as patch-wise graphs using CLIP-derived features. By applying graph matching, GSS captures spatial and semantic relationships beyond pixel-level similarities. Our approach aligns with neuroscientific models of visual processing and demonstrates robust, interpretable results that complement existing metrics.

In this work, we present a Deep Learning framework to predict the progenitor star’s characteristics of Supernovae (SNe) from their observed light curves. This task is crucial for astrophysics, as it can provide insights into the evolution of the star before the explosion and into the SN explosion mechanism. In fact, there is no direct mapping between the observed light curves and the progenitor’s characteristics, and the common techniques used to infer them are indirect, meaning that they rely on the comparison between the observed light curves and the light curves generated with some physical model simulation using the supposed progenitor’s characteristics as input. However, the physical models used to generate the light curves are not perfect, being either computationally expensive or based on simplifying assumptions. Here, we train a machine learning model on a dataset of light curves generated with a semi-analytical model — which is computationally efficient and accurate for the problem at hand — to predict the progenitor’s mass, radius, energy, and nickel mass from the observed light curves of SNe similar to SN 1987A. Our results show that the Deep Learning model effectively learns the complex mapping between the observed light curves and the progenitor’s characteristics with a low mean absolute percentage error, and we note that the proposed framework is general and can be applied to other types of SNe without significant modifications.

Centrality is a relevant topic in the field of network research, due to its various theoretical and practical implications. In general, all centrality metrics aim at measuring the importance of nodes (according to some definition of importance), and such importance scores are used to rank the nodes in the network, therefore the rank improvement is a strictly related topic. In a given network, the rank improvement is achieved by establishing new links, therefore the question shifts to which and how many links should be collected to get a desired rank. This problem, also known as link-building has been shown to be NP-hard, and most heuristics developed failed in obtaining good performance with acceptable computational complexity. In this article, we present LB–GDM , a novel approach that leverages Geometric Deep Learning to tackle the link-building problem. To validate our proposal, 31 real-world networks were considered; tests show that LB–GDM performs significantly better than the state-of-the-art heuristics, while having a comparable or even lower computational complexity, which allows it to scale well even to large networks.

Logistic optimization is a strategic element in many industrial processes, given that an optimized logistics makes the processes more efficient.A relevant case, in which the optimization of logistics can be decisive, is the maintenance in a Wind Farm where it can lead directly to a saving of energy cost.Wind farm maintenance presents, in fact, significant logistical challenges.They are usually distributed throughout the territory and also located at considerable distances from each other, they are generally found in places far from uninhabited centers and sometimes difficult to reach and finally spare parts are rarely available on the site of the plant itself.In this paper, we will study the problem concerning the optimization of maintenance logistics of wind plants based on the use of specific vehicle routing optimization algorithms.In particular a pickup-anddelivery algorithm with time-window is adopted to satisfy the maintenance requests of these plants, reducing their management costs.The solution was applied to a case study in a renewable energy power plant.Results time reduction and simplification and optimization obtained in the real case are discussed to evaluate the effectiveness and efficiency of the adopted approach.

The provision of wellness in workplaces gained interest in the last decades. A factor that contributes significantly to workers' health is their diet, expecially when provided by canteen services. The assessment of such a service involves questions as food cost, its sustainability, quality, nutritional facts and variety, as well as employees' health and diseases prevention, productivity increase, economic convenience vs eating satisfaction when using canteen services. In this paper, a multi-company canteen service dataset is presented and first significant considerations, as well as future directions, are discussed.

Internet of things can be viewed as the shifting from a network of computers to a network of things.To support M2M communication, several protocols have been developed; many of them are endorsed by client-broker model with a publish-subscribe interaction mechanism. In this paper we introduce a multi broker solution where the network of brokers is inspired by social relationships. This allow data sharing among several IoT systems, leads to a reliable and effective query forwarding algorithm and the small world effect coming from mimic humans relations guarantees fast responses and good query recall.

The number and type of digital sources storing healthcare data is increasing more and more, rising the problem of collecting actually dispersed information about a single patient.In this paper we propose an agent-based system to support integration of health-related data extracted from both structured (HIS) and semi-structured (websites and social networks) sources.Integrated data are exported in HL7 format to finally feed personal health record (PHR).

SUMMARY Currently several computer‐based scenarios leverage the concept of trust as a mean to make electronic interactions (e.g., e‐commerce transactions) as reliable as possible, allowing to cope with uncertainty and risks by recommending trusted peers. Generally, the evaluation of trustworthiness can be accomplished according to many principles, from social‐based to psychology‐based; one of the commonly adopted approaches within peer‐to‐peer networks, virtual social networks, and recommendation systems is the reputation‐based trust evaluation. Because more and more large networks (even with millions of nodes) aim at leveraging trust, approaches to its assessment have to take into account the factors as efficient distributed implementation and effective security protection against malicious attacks. In this paper, we present a distributed and secure algorithm based on TrustWebRank, a metric that takes into account both personalized trust evaluation and network dynamics issues. To test our proposal both in terms of complexity and bandwidth usage, we performed simulations on a large and real dataset built from the Epinions.com recommendation system. Results show that the proposed distributed algorithm is effective and efficient, while preserving original benefits of TrustWebRank. Copyright © 2011 John Wiley & Sons, Ltd.

The use of P2P approach in e-learning is an interesting solution in particular within lifelong learning. Generally, peer-to-peer paradigm is a successful solution to the problem of resources sharing. In this field there are two open questions. First, how to advertise learning objects on the network, so that peers knows they exist. Second, how to choose the best set of learning objects. The system presented in this paper aims at answering these questions.

P2P communication paradigm is a successful solution to the problem of resources sharing as shown by the numerous real overlay networks present on Internet. One of the issue of P2P networks is how a resource shared by a peer can be made known to the other peers or, in other words, how to advertise a resource on the network. In this paper we propose an advertising method for PROSA, a P2P architecture inspired by social relationships. In the paper we show that the introduction of our resources advertising method improves PROSA performance with a low overhead.

The e-learning represents the new frontier of education, significantly improving the learning process. We propose an e-learning model, providing both teachers and students with an open and modular learning environment. We then focus on courses personalization, both in terms of contents and teaching materials, according to each student's needs and capabilities, also taking teacher guidelines into account. To accomplish this, we model courses/lessons as graph nodes, where arcs represent their precedence/succession relationships. We outline a course generation/presentation engine which allows the creation of personalized learning paths (subgraph) by extracting lessons, eliminating those known to the student, and arranging them into a tree including all possible paths starting from the student's possessed knowledge towards desired knowledge.

Traditional servers offer little support to handle situations like faults due to the approach usually followed based on mirroring and/or duplicating functions. This approach lacks in transparency with respect to the client and causes a considerable waste of resources. Some requirements of a good service provider have been highlighted such as, for instance, reliability, capability to work in a heterogeneous environment, scalability and dynamical reconfigurability. A service provider has been developed in order to satisfy the above mentioned requirements; it is based mainly on the use of multicast addressing and the agent paradigm.

CoDesign aims to integrate the design techniques of hardware and software. In this work, we present a CoDesign methodology based on a formal approach to embedded system specification. This methodology uses the Templated T-LOTOS language to specify the system during all design phases. Templated T-LOTOS is a formal language based on CCS and CSP models. Using Templated T-LOTOS, a system can be specified by observing the temporal ordering in which the events occur from the outside. In this paper we focus on the synthesis of system specified by Templated T-LOTOS. The proposed synthesis algorithm takes advantage of peculiarities of Templates T-LOTOS. Hardware modules are translated into a register transfer-level language that manages some signals in order to drive synchronization, while the software models are translated into C according to a finite state model whose operations are controlled by a scheduler. The synthesis of the Templated T-LOTOS specification is based on the direct translation of the language operators to ensure that the implemented system is the same as the specified one.