Abstract
Missing data is an issue often addressed with imputation strategies that replace the missing values with plausible ones. A trend in these strategies is the use of generative models, one being Variational Autoencoders. However, the default loss function of this method gives the same importance to all data, while a more suitable solution should focus on the missing values. In this work an extension of this method with a custom loss function is introduced (Variational Autoencoder with Weighted Loss). The method was compared with state-of-the-art generative models and the results showed improvements higher than 40% in several settings. © ESANN 2020 - Proceedings, 28th European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning.

Abstract

Abstract
Conducting epidemiological and risk assessment research that considers the exposome concept, as in the case of HEALS project, requires the acquisition of higher dimension data sets of an increased complexity. In this context, new methods that provide accurate and interpretable data summary on relevant environmental factors are of major importance. In this work, a questionnaire was developed to collect harmonized data on potential pollutant sources to air in the indoor environment where children spend an important part of their early life. The questionnaire was designed in a user friendly checklist format to be filled out at the maternity in ten European cities. This paper presents and discusses the rationale for the selection of the questionnaire contents and the results obtained from its application in the households of 309 HEALS-enrolled families with babies recently born in Porto, Portugal. The tool was very effective in providing data on the putative air pollution sources in homes, with special focus on the bedroom of the newborns. The data collected is part of a wider effort to build the databases and risk assessment models of the HEALS project. The results of the analysis of the collected data suggest that, for the population under study, the main concerns on early life exposures at home can be related to emissions from the use of household solid fuels, indoor tobacco, household cleaning products, fragranced consumer products (e.g. air fresheners, incense and candles), moisture-related pathologies and traffic-related outdoor pollution. Furthermore, it is anticipated that the tool can be a valuable means to empower citizens to actively participate in the control of their own exposures at home. Within this context, the application of the checklist will also allow local stakeholders to identify buildings presenting most evident IAQ problems for sampling or intervention as well as to guide them in preparing evidence-based educational/awareness campaigns to promote public health through creating healthy households.

Abstract
The use of optical sensors inside the needle can improve targeting precision and can bring real-time information about the location of the needle tip if necessary, since a needle bends through insertion into the tissue. Therefore, the precise location of the needle tip is so important in percutaneous treatments. In the current experiment, a fiber sensor based on a Fabry-Perot (FP) cavity is described to measure the needle curvature. The sensor is fabricated by producing an air bubble between two sections of multimode fiber. The needle with the sensor therein was attached at one end and deformed by the application of movements. The sensor presents a sensitivity of -0.152 dB/m(-1) to the curvature measurements, with a resolution of 0.089 m(-1). The sensory structure revealed to be stable, obtaining a cross-sensitivity to be 0.03 m(-1)/degrees C.

Abstract
This paper proposes a comprehensive framework for generation and transmission planning of renewable dominated power systems, which is formulated as a stochastic multi-objective problem. In this regard, a Normalized Normal Constraint (NNC) solution approach is proposed to solve the introduced stochastic multiobjective generation and transmission planning (GTP) problem. The NNC is utilized in this paper as a relation between different objective functions with different dimensions to find the optimal weighting factors of these objectives. The NNC is applied for solving the GTP problem with objective functions including the investment and operation costs along with the transmission losses, while considering the cost of unserved energy, as well as the uncertainty of load and Renewable Energy Resources (RERs). A fuzzy-based decision making framework is utilized to select the best solution among the optimal non-dominated solution points. A scenario-based approach is used to model the uncertainties. The Garver 6-bus and IEEE 118-bus test systems are utilized to perform the numerical analysis. The simulation results validate the performance and importance of the proposed model, as well as the effectiveness of the NNC to find the evenly distributed Pareto solutions of the multiobjective problems.

Abstract
The loss of rotational inertia in the future power systems motivate the need to find new technical solutions for this challenge. Many solution have emerged in last years based on power converters with emulation of inertia. In this paper, we explore the concept of distributed virtual inertia (DVI) by investigating a new control method that use the common DC bus capacitors in the DC link of the converter as a buffer of energy to provide an equivalent mechanical inertial response. Using time-domain simulations with a detailed two-stage Photovoltaic (PV) inverter, we observe and discuss the benefits in the grid frequency.