Multi-scale stacked sequential learning

Sequential learning is the discipline of machine learning that deals with dependent data such that neighboring labels exhibit some kind of relationship. The paper main contribution is two-fold: first, we generalize the stacked sequential learning, highlighting the key role of neighboring interactions modeling. Second, we propose an effective and efficient way of capturing and exploiting sequential correlations that takes into account long-range interactions. We tested the method on two tasks: text lines classification and image pixel classification. Results on these tasks clearly show that our approach outperforms the standard stacked sequential learning as well as state-of-the-art conditional random fields.     

An Attempt at the Classification of Energy Decaying Schemes for Structural and Multibody Dynamics

This paper analyzes the formulation of energy preserving/decaying schemes for dynamics problems. We argue that any energy preserving/decaying scheme can always be seen as composed of an underlying temporal discretization, that is then slightly modified in order to prove a discrete energy bound within a time step. While the details of the modified scheme depend in a critical way on the governing equations, the underlying discretization can in principle be applied to a variety of models. We review some of the temporal underlying schemes recently proposed in the literature, presenting them with a common notation. We show their similarities and highlight their differences.     

Optimum design of linear tuned mass dampers for structures with nonlinear behaviour

This paper investigates the optimum design of tuned mass dampers (TMDs) for the seismic protection of inelastic structures. A single linear TMD is treated and is assumed to be applied to a single nonlinear degree of freedom system, described by the Bouc–Wen hysteretic model. The seismic load is modelled by a stationary filtered stochastic process to consider its intrinsic stochastic nature. The optimization problem is set by taking into consideration three different possible objective functions (OFs): the maximum of the peak structural displacement standard deviation, the average hysteretic dissipated energy of a protected building with reference to an unprotected one, and a functional damage that considers the two indexes previously described. Different numerical examples and parametric analysis are shown to confront the three optimization criteria and to determine the best tuning frequency and damping ratio of the TMDs to be used in the structure. Results confirm that the application of a TMD system reduces the amount of the hysteretic dissipated energy, which is a direct measure of damage in the structure, and so it is beneficial for the protection of buildings that develop a nonlinear behaviour under severe dynamic loadings.     

Engineering the Losses and Beam Divergence in Arrays of Patch Antenna Microcavities for Terahertz Sources

We perform a comprehensive study on the emission from finite arrays of patch antenna microcavities designed for the terahertz range by using a finite element method. The emission properties including quality factors, far-field pattern, and photon extraction efficiency are investigated for etched and non-etched structures as a function of the number of resonators, the dielectric layer thickness, and period of the array. In addition, the simulations are achieved for lossy and perfect metals and dielectric layers, allowing to extract the radiative and non-radiative contributions to the total quality factors of the arrays. Our study shows that this structure can be optimized to obtain low beam divergence (FWHM <10°) and photon extraction efficiencies >50% while keeping a strongly localized mode. These results show that the use of these microcavities would lead to efficient terahertz emitters with a low divergence vertical emission and engineered losses.     

Magnetically controllable gastrointestinal steering of video capsules

Wireless capsule endoscopy (WCE) allows for comfortable video explorations of the gastrointestinal (GI) tract, with special indication for the small bowel. In the other segments of the GI tract also accessible to probe gastroscopy and colonscopy, WCE still exhibits poorer diagnostic efficacy. Its main drawback is the impossibility of controlling the capsule movement, which is randomly driven by peristalsis and gravity. To solve this problem, magnetic maneuvering has recently become a thrust research area. Here, we report the first demonstration of accurate robotic steering and noninvasive 3-D localization of a magnetically enabled sample of the most common video capsule (PillCam, Given Imaging Ltd, Israel) within each of the main regions of the GI tract (esophagus, stomach, small bowel, and colon) in vivo, in a domestic pig model. Moreover, we demonstrate how this is readily achievable with a robotic magnetic navigation system (Niobe, Stereotaxis, Inc, USA) already used for cardiovascular clinical procedures. The capsule was freely and safely moved with omnidirectional steering accuracy of 1°, and was tracked in real time through fluoroscopic imaging, which also allowed for 3-D localization with an error of 1 mm. The accuracy of steering and localization enabled by the Stereotaxis system and its clinical accessibility world wide may allow for immediate and broad usage in this new application. This anticipates magnetically steerable WCE as a near-term reality. The instrumentation should be used with the next generations of video capsules, intrinsically magnetic and capable of real-time optical-image visualization, which are expected to reach the market soon.     

From the Far Side of the Moon： Delay/Disruption- Tolerant Networking Communications via Lunar Satellites

Delay/Disruption-Tolerant Network- ing （DTN） originated from research on Interplanetary Internet and still today space appli- cations are the most important application field and research stimulus. This paper investigates DTN communications between the Earth and the far side of the Moon, by means of a lunar orbiter acting as relay. After an introductory part, the paper presents a comprehensive ana- lysis of the DTN performance that can be ac- hieved on the identified communication scen- ario. The focus is on the evaluation of the state- of-the-art ability of Interplanetary Overlay Network （ION）, the NASA DTN implementation of Bundle Protocol （BP） and Contact Graph Routing （CGR）, to meet the many challenges of the space communication scenario investigated （and more generally of a future interplay- netary Internet）： intermittent links, network partitioning, scarce bandwidth, long delays, dyn- amic routing, handling of high priority and emergency traffic, interoperability issues. A study of security threats and Bundle Security Protocol （BSP） countermeasures complete the work. The many results provided, confirm the essential role of DTN in future space communications.     

Carbon nanotubes/polydimethylsiloxanes systems for thermal management of miniaturized electronic components

The thermal performances of nanocomposite layers formed by Single-walled Carbon Nanotubes (SWCNT) dispersed in 2 different kind of polydimethyl-siloxane (PDMSO) matrices has been investigated by measuring the thermal resistance under conditions similar to the ones used for thermal management in microelectronics. A series of nanocomposite samples with thickness in the range 25 microm(-1) cm have been tested. The nanocomposites were prepared varying the amounts of nanotubes embedded in the matrix (from 0.1 to 5%w). In some cases also microsized graphites were mixed to the nanotube's fillers. For 25 micron thick layers, the thermal resistance of the neat silicone specimen can be reduced of 54% with the addition of 2%w carbon nanotubes. The variation of thermal conductivity as a function of the SWCNT's loading is reported and discussed. Furthermore the dispersion's effects of the nanotubes in the layers and the effects on the realization of a net-like system have been investigated.     

Effect of sourdough fermentation on stabilisation,and chemical and nutritional characteristics of wheat germ

Lactic acid bacteria strains were identified from wheat germ by 16S rRNA partial sequencing, subjected to RAPD-PCR typing and screened. Lactobacillus plantarum LB1 and Lactobacillus rossiae LB5 were used as starters to produce sourdough fermented wheat germ (SFWG). The chemical and nutritional characteristics of SFWG were compared to those of the raw wheat germ (RWG). Lipase activity in SFWG was ca. 2.6-fold lower than that found in RWG. As shown by SPME/GC/MS analysis, most of the volatile compounds derived from lipid oxidation during storage (40 days) were at markedly lower levels in SFWG compared to RWG. Fermentation of wheat germ increased of ca. 50% the concentration of free amino acids. Glu markedly decreased in SFWG, due to its conversion in GABA. The concentration of the anti-nutritional factor raffinose also decreased in SFWG. The in vitro protein digestibility, the concentration of total phenols, phytase and antioxidant activities were increased by fermentation.