Abstract
In time series analysis, Autoregressive Moving Average (ARMA) models play a central role. Because of the importance of parameter estimation in ARMA modeling and since it is based on aggregate time series so often, we analyze the effect of temporal aggregation on estimation accuracy. We derive the relationships between the aggregate and the basic parameters and compute the actual values of the former from those of the latter in order to measure and compare their estimation accuracy. We run a simulation experiment that shows that aggregation seriously worsens estimation accuracy and that the impact increases with the order of aggregation.

Abstract
Increasing penetration of distributed generation (DG), may be interesting from several points of view, but it raises important challenges about distribution system operation and planning practices. To optimal allocation of DG, which play an important role in construction of microgrids, the benefits and risks should be qualified and quantified. This paper introduces several probabilistic indices to evaluate the potential operational effects of increasing penetration of renewable DG units such as wind power and photovoltaic on rural distribution network with the aid of evaluating technical benefits and risks tradeoffs. A probabilistic generation-load model is suggested to calculate these indices which combine a large number of possible operating conditions of renewable DG units with their probabilities. Temporal and annual indices of voltage profile and line flow-related attributes such as interest voltage rise, risky voltage rise, risky voltage down, line loss reduction, line loss increment, and line overload flow are introduced using probability and expected values of their occurrence. Also, to measure the overall interests and risks of installing DG, composite indices are presented. The implementation of the proposed framework in a 4-bus and IEEE 33-bus radial distribution systems shows the effectiveness of the benefits and risks assessment technique with the proposed metrics.

Abstract
This paper proposes a predictive dispatch model to manage energy flexibility in the domestic energy system. Electric Vehicles (EV), batteries and shiftable loads are devices that provide energy flexibility in the proposed system. The proposed energy management problem consists of two stages: day-Ahead and real time. A hybrid method is defined for the first time in this paper to model the uncertainty of the PV power generation based on its power prediction. In the day-Ahead stage, the uncertainty is modeled by interval bands. On the other hand, the uncertainty of PV power generation is modeled through a stochastic scenario-based method in the real-Time stage. The performance of the proposed hybrid Interval-Stochastic (InterStoch) method is compared with the Modified Stochastic Predicted Band (MSPB) method. Moreover, the impacts of energy flexibility and the demand response program on the expected profit and transacted electrical energy of the system are assessed in the case study presented in this paper.

Abstract
The exploration of the physical layer characteristics of the wireless channel is currently the object of intensive research in order to develop advanced secrecy schemes that can protect information against eavesdropping attacks. Following this line of work, in this manuscript we consider a massive MIMO system and jointly design the channel precoder and security scheme. By doing that we ensure that the precoding operation does not reduce the degree of secrecy provided by the security scheme. The fundamental working principle of the proposed technique is to apply selective random rotations in the transmitted signal at the antenna level in order to achieve a compromise between legitimate and eavesdropper channel capacities. These rotations use the phase of the reciprocal wireless channel as a common random source between the transmitter and the intended receiver. To assess the security performance, the proposed joint scheme is compared with a recently proposed approach for massive MIMO systems. The results show that, with the proposed joint design, the number of antenna elements does not influence the eavesdropper channel capacity, which is proved to be equal to zero, in contrast to previous approaches.

Abstract
More and more cyber-physical systems and the internet of things push for a multitude of devices and systems, which need to work together to provide the services as required by the users. Nevertheless, the speed of development and the heterogeneity of devices introduces considerable challenges in the development of such systems. This paper describes a solution being implemented in the setting of a serious game scenario, connected to real homes energy consumption. The solution provides a publish-subscribe middleware which is able to seamlessly connect all the components of the system.

Abstract
Cancer genome projects are characterizing the genome, epigenome and transcriptome of a large number of samples using the latest high-throughput sequencing assays. The generated data sets pose several challenges for traditional statistical and machine learning methods. In this work we are interested in the task of deriving the most informative genes from a cancer gene expression data set. For that goal we built denoising autoencoders (DAE) and stacked denoising autoencoders and we studied the influence of the input nodes on the final representation of the DAE. We have also compared these deep learning approaches with other existing approaches. Our study is divided into two main tasks. First, we built and compared the performance of several feature extraction methods as well as data sampling methods using classifiers that were able to distinguish the samples of thyroid cancer patients from samples of healthy persons. In the second task, we have investigated the possibility of building comprehensible descriptions of gene expression data by using Denoising Autoencoders and Stacked Denoising Autoencoders as feature extraction methods. After extracting information related to the description built by the network, namely the connection weights, we devised post-processing techniques to extract comprehensible and biologically meaningful descriptions out of the constructed models. We have been able to build high accuracy models to discriminate thyroid cancer from healthy patients but the extraction of comprehensible models is still very limited.