Abstract
Deep transfer learning emerged as a new paradigm in machine learning in which a deep model is trained on a source task and the knowledge acquired is then totally or partially transferred to help in solving a target task. In this paper, we apply the source-target-source methodology, both in its original form and an extended multi-source version, to the problem of cross-sensor biometric recognition. We tested the proposed methodology on the publicly available CSIP image database, achieving state-of-the-art results in a wide variety of cross-sensor scenarios.

Abstract
The increase penetration of renewable energy sources (RES) into the European power system has introduced a significant amount of variability and uncertainty in the generation profiles raising the needs for ancillary services as well as other tools like demand response, improved generation forecasting techniques and changes to the market design. While RES is able to replace energy produced by the traditional centralized generation, it cannot displace its capacity in terms of ancillary services provided. Therefore, centralized generation capacity must be retained to perform this function leading to over-capacity issues and underutilisation of the assets. Large-scale reversible hydro power plants represent the majority of the storage solution installed in the power system. This technology comes with high investments costs, hence the constant search for methods to increase and diversify the sources of revenue. Traditional fixed speed pump storage units typically operate in the day-ahead market to perform price arbitrage and, in some specific cases, provide downward replacement reserve (RR). Variable speed pump storage can not only participate in RR but also contribute to FRR, given their ability to control its operating point in pumping mode. This work does an extended analysis of a complete bidding strategy for Pumped Storage Power, enhancing the economic advantages of variable speed pump units in comparison with fixed ones.

Abstract

Abstract
The capability of an Autonomous Underwater Vehicle (AUV) or Remote Operated Vehicles (ROVs) to communicate with underwater sensor nodes or a docking station, for the exchange or transfer of data gathered during a survey mission, requires an high-speed short-range communication link. This is important because of the global attention to the underwater communication applications. To this end, underwater antennas will play a significant role in ensuring good data rates and propagation distances for these applications. In this paper, the performance of three antennas, specifically, loop, dipole and J-pole is assessed through simulation for usage in fresh and sea water operating in the High Frequency (HF) band. The antennas were designed in FEKO, an electromagnetic simulation software and their performance is assessed in terms of bandwidth and directivity. The results obtained shows that the J-pole antenna has significant advantages in term of the measured parameters over the other antennas. Experimental results of the reflection coefficient of the J-pole antenna in fresh water are given that agree well with the simulation results. © 2017 IEEE.

Abstract
Control and monitoring systems capabilities are unavoidable to improve product quality, reduce production time, and to the rapid adaptation to changes in production. Therefore, it is advantageous to develop control and monitoring systems, and maximize the use of the resources already available in machine tools, without high additional costs and hard implementations. The use of Programmable Logic Controllers (PLC) on industries is growing. This sort of controller has been an election tool to attend present day requirements. It has the necessary resources to acquire and manage the information and control of the machines, and easily interacts with SCADA systems. The objective of the study described in this paper was to develop a solution for monitoring a metal casting machine, which could replace an outdated system installed on it, and find a solution to control a metal pouring process. All programs that allow the PLC to control the machine movements, and also perform the casting process, were developed during the project implementation. The results show that the developed solution is able to control the machine without the need to invest in a PC, which is a more expensive solution and with a more limited life time. © Springer International Publishing Switzerland 2017.

Abstract
Heart sounds are difficult to interpret because a) they are composed by several different sounds, all contained in very tight time windows; b) they vary from physiognomy even if the show similar characteristics; c) human ears are not naturally trained to recognize heart sounds. Computer assisted decision systems may help but they require robust signal processing algorithms. In this paper, we use a real life dataset in order to compare the performance of a hidden Markov model and several hidden semi Markov models that used the Poisson, Gaussian, Gamma distributions, as well as a non-parametric probability mass function to model the sojourn time. Using a subject dependent approach, a model that uses the Poisson distribution as an approximation for the sojourn time is shown to outperform all other models. This model was able to recreate the "true" state sequence with a positive predictability per state of 96%. Finally, we used a conditional distribution in order to compute the confidence of our classifications. By using the proposed confidence metric, we were able to identify wrong classifications and boost our system (in average) from an approximate to 83% up to approximate to 90% of positive predictability per sample.