Attention is an increasingly popular mechanism used in a wide range of neural architectures. The mechanism itself has been realized in a variety of formats. However, because of the fast-paced advances in this domain, a systematic overview of attention is still missing. In this article, we define a unified model for attention architectures in natural language processing, with a focus on those designed to work with vector representations of the textual data. We propose a taxonomy of attention models according to four dimensions: the representation of the input, the compatibility function, the distribution function, and the multiplicity of the input and/or output. We present the examples of how prior information can be exploited in attention models and discuss ongoing research efforts and open challenges in the area, providing the first extensive categorization of the vast body of literature in this exciting domain.

In this work, a flexible and extensive digital platform for Smart Homes is presented, exploiting the most advanced technologies of the Internet of Things, such as Radio Frequency Identification, wearable electronics, Wireless Sensor Networks, and Artificial Intelligence. Thus, the main novelty of the paper is the system-level description of the platform flexibility allowing the interoperability of different smart devices. This research was developed within the framework of the operative project HABITAT (Home Assistance Based on the Internet of Things for the Autonomy of Everybody), aiming at developing smart devices to support elderly people both in their own houses and in retirement homes, and embedding them in everyday life objects, thus reducing the expenses for healthcare due to the lower need for personal assistance, and providing a better life quality to the elderly users.

Permanganate coatings show promise as replacements for chromate conversion coatings on aluminum alloys used in the aerospace industry. The development of permanganate conversion coatings on AA2024-T3 (UNS A92024) from acid solution was followed for up to 240 s using scanning electron microscopy (SEM), x-ray photoelectron spectroscopy (XPS), Auger electron spectroscopy (AES), and Rutherford backscattering spectroscopy (RBS). The permanganate coatings have a manganese dioxide (MnO2)-rich exterior surface covering a mixed Al-Mn oxide with a composition close to MnO2·Al2O3. RBS and AES depth profiles indicate that Mn is distributed throughout the oxide over the metal matrix, while XPS confirmed that Mn(IV) and Al(III) were both present in the coating. The onset of coating formation from permanganate solutions is rapid (<30 s) with the maximum thickness (50 nm to 70 nm) being reached after approximately 60 s. The limit on coating thickness may be due to the passivation of boehmite (AlOOH) by MnO2. Increased deoxidation time prior to coating had little effect on the final coating thickness, but did increase the degree of manganese incorporation.

We explore the use of residual networks and neural attention for multiple argument mining tasks. We propose a residual architecture that exploits attention, multi-task learning, and makes use of ensemble, without any assumption on document or argument structure. We present an extensive experimental evaluation on five different corpora of user-generated comments, scientific publications, and persuasive essays. Our results show that our approach is a strong competitor against state-of-the-art architectures with a higher computational footprint or corpus-specific design, representing an interesting compromise between generality, performance accuracy and reduced model size.

In the Pre-Alps and Alps of north-eastern Italy, where the First World War was fought, the discovery of human remains is not a rare event. In the past year, the skeleton of an Italian soldier was f...