Neural networks for variational problems in engineering

In this work a conceptual theory of neural networks (NNs) from the perspective of functional analysis and variational calculus is presented. Within this formulation, the learning problem for the multilayer perceptron lies in terms of finding a function, which is an extremal for some functional. Therefore, a variational formulation for NNs provides a direct method for the solution of variational problems. This proposed method is then applied to distinct types of engineering problems. In particular a shape design, an optimal control and an inverse problem are considered. The selected examples can be solved analytically, which enables a fair comparison with the NN results. Copyright © 2008 John Wiley & Sons, Ltd.     

Comparison of system performance at 50, 62.5 and 100 GHz channel spacing over transoceanic distances at 40 Gbit/s channel rate using RZ-DPSK

Assuming RZ-DPSK format at 40 Gbit/s channel rate, the impact on the Q-factor of a variable channel spacing has been evaluated in five WDM experiments conducted in a recirculating loop over transoceanic distances. A degradation of the Q-factor by nearly 3 dB is observed when the channel spacing is reduced from 100 to 50 GHz     

Reducing hysteresis in magnetostrictive-piezoelectric magneticsensors

In this paper, we present a method to avoid hysteresis in magnetostrictive-piezoelectric sensors. It is based on the application of an alternating magnetic field of a frequency different than the piezoelectric exciting frequency that supplies enough energy to overcome any domain wall pinning. The signal is filtered with a notch filter synchronized to the magnetic field frequency and thus it does not contribute to the useful response of the sensor. The results show an appreciable decrease in the magnetic hysteresis and also a small decrease in the sensitivity of the sensor     

Searching for Lyme disease in Colombia: a preliminary study on the vector

Since ageing is accompanied by various alterations of the immune system, the aim of the present study was to investigate the effect of age on the expression of surface markers in peripheral blood monocytes. We studied 28 healthy young subjects and 28 elderly subjects fulfilling the criteria of the SENIEUR protocol. Peripheral blood mononuclear cells were isolated and the expression of various surface markers was analysed by double colour flow cytometry. The mean fluorescence intensity of the intercellular adhesion molecule- (CD54), CD29, CD45RO and CD32 was increased significantly in CD14+ cells of elderly people when compared with the young subjects. No significant differences were found in the expression of CD11a, CD11b, CD15, CD26, CD27, CD33, CD45RA, CD45RB, CD49d, HLA-DR and CD65. In summary, we demonstrated significant increase in four monocyte surface markers in subjects of old age; this finding may be related to certain immune phenomena observed in ageing subjects such as auto-immunity.     

Differential colocalization of estrogen receptor beta (ERbeta) with oxytocin and vasopressin in the paraventricular and supraoptic nuclei of the female rat brain: an immunocytochemical study

Evidence exists for the localization of the newly identified estrogen receptor beta (ERbeta) within the rat paraventricular nucleus (PVN) and supraoptic nucleus (SON), regions which lack ERalpha. Presently, we investigate whether ERbeta-like-immunoreactivity (-ir) is found within cells of several major neuropeptide systems of these regions. Young adult Sprague-Dawley rats were ovariectomized (OVX), and 1 week later half of the animals received estradiol-17beta (E). Dual-label immunocytochemistry was performed on adjacent sections by using an ERbeta antibody, followed by an antibody to either oxytocin (OT), arginine-vasopressin (AVP), or corticotropin releasing hormone. Nuclear ERbeta-ir was identified within SON and retrochiasmatic SON, and in specific PVN subnuclei: medial parvicellular part, ventral and dorsal zones, dorsal and lateral parvicellular parts, and in the posterior magnocellular part, medial and lateral zones. However, the ERbeta-ir within magnocellular areas was noticeably less intense. OT-/ERbeta-ir colocalization was confirmed in neurons of the parvicellular subnuclei, in both OVX and OVX+E brains ( approximately 50% of OT and 25% of ERbeta-labeled cells between bregma -1.78 and -2.00). In contrast, few PVN parvicellular neurons contained both AVP- and ERbeta-ir. As well, very little overlap was observed in the distribution of cells containing corticotropin releasing hormone- or ERbeta-ir. In the SON, most nuclear ERbeta-ir colocalized with AVP-ir, whereas few OT-/ERbeta-ir dual-labeled cells were observed. These findings suggest that estrogen can directly modulate specific OT and AVP systems through an ERbeta-mediated mechanism, in a tissue-specific manner.     

The central pixel of the MAGIC telescope for optical observations

The MAGIC telescope has been designed for the observation of Cherenkov light generated in Extensive Air Showers initiated by cosmic particles. However, its 17 m diameter mirror and optical design makes the telescope suitable for direct optical observations as well. In this paper, we report about the development of a system based on the use of a dedicated photo-multiplier (PMT) for optical observations. This PMT is installed in the centre of the MAGIC camera (the so-called central pixel). An electro-to-optical system has been developed in order to transmit the PMT output signal by an optical fibre to the counting room, where it is digitized and stored for off-line analysis. The performance of the system using the optical pulsation of the Crab nebula as calibration source is presented. The time required for a 5σ detection of the Crab pulsar in the optical band is less than 20 s. The central pixel will be mainly used to perform simultaneous observations of the Crab pulsar both in the optical and γ-ray regimes. It will also allow for periodic testing of the precision of the MAGIC timing system using the Crab rotational optical pulses as a very precise timing reference.     

Cost-adaptive motion estimation strategy for high-performance video encoders

An adaptive cost block matching (ACBM) motion estimation technique is proposed to achieve a good compromise between the visual quality of decoded video sequences and the computational cost. Simulation results demonstrate that the ACBM algorithm is able to obtain a better rate-distortion performance that the one given by the full-search algorithm, with reductions of up to 90% in the computational cost.     

Determination of mass transport properties in food/packaging systems by local measurement with Raman microspectroscopy

A fast, nondestructive method based on the determination of local concentration profiles in the polymer thickness with Raman microspectroscopy is presented here. It was used to assess the diffusivity of a model molecule (p‐terphenyl) in amorphous polystyrene films at 95°C (2.38 ± 1.08 × 10^?17 m²/s). This methodology was validated by comparison with a more classical destructive approach based on the monitoring of the concentration evolution in the whole of the film with gas chromatography (89.4 × 10^?17 m²/s). These values were in agreement with data available in the literature for molecules of the same molecular weight and temperature range determined with local measurement and were significantly lower than those determined by global measurement. Raman microspectroscopy was found to be adapted to slow diffusion speeds typically found in high‐barrier polymers; this allowed diffusivity to be obtained long before the equilibrium was reached and, thus, without the need for the partition coefficient. © 2014 Wiley Periodicals, Inc. J. Appl. Polym. Sci. 2014 , 131, 40958.     

Removing redundant TCP functionalities in wired‐cum‐wireless networks with IEEE 802.11e HCCA support

The TCP was originally designed for wired networks, assuming transmission errors were negligible. Actually, any acknowledgment time‐out unconditionally triggers the congestion control mechanism, even in wireless networks in which this assumption is not valid. Consequently, in wireless networks, TCP performance significantly degrades. To avoid this degradation, this paper proposes the so‐called split TCP and UDP. In this approach, the access point splits the TCP connection and uses a customized and lighter transport protocol for the wireless segment. It takes advantage of the IEEE 802.11e Hybrid Coordination Function Controlled Channel Access (HCCA) mechanisms to remove redundant TCP functionalities. Specifically, the HCCA scheduler allows disabling of the congestion control in the wireless link. Similarly, the IEEE 802.11e error control service makes possible to eliminate TCP acknowledgments, therefore reducing the TCP protocol overhead. Finally, the usage of an HCCA scheduler permits providing fairness among the different data flows. The proposed split scheme is evaluated via extensive simulations. Results show that split TCP and User Datagram Protocol outperforms the analyzed TCP flavors—specifically designed for wireless environments—and the split TCP solution, achieving up to 95% of end‐user throughput gain. Furthermore, the proposed solution is TCP friendly because TCP flows are not degraded by the presence of flows by using this approach. Copyright © 2013 John Wiley & Sons, Ltd.