Abstract
Traditionally, a single model is developed for a data mining task. As more data is being collected at a more detailed level, organizations are becoming more interested in having specific models for distinct parts of data (e. g. customer segments). From the business perspective, data can be divided naturally into different dimensions. Each of these dimensions is usually hierarchically organized (e. g. country, city, zip code), which means that, when developing a model for a given part of the problem (e. g. a zip code) the training data may be collected at different levels of this nested hierarchy (e. g. the same zip code, the city and the country it is located in). Selecting different levels of granularity may change the performance of the whole process, so the question is which level to use for a given part. We propose a metalearning model which recommends a level of granularity for the training data to learn the model that is expected to obtain the best performance. We apply decision tree and random forest algorithms for metalearning. At the base level, our experiment uses results obtained by outlier detection methods on the problem of detecting errors in foreign trade transactions. The results show that using metalearning help finding the best level of granularity.

Abstract
Self-localization of a robot in an indoor plant is one of the most important requirement in mobile robotics. This paper addresses the application and improvement of a well known localization algorithm used in Robocup Midsize league competition in real service and industrial robots. This new robust approach is based on modeling the quality of several measures and minimizing the maching error. The presented innovative work applies the robotic football knowledge to other fields with high accuracy. Real and simulated results allow to validate the proposed methodology.

Abstract
MATLAB to C translation is foreseen to raise the overall abstraction level when mapping computations to embedded systems (possibly consisting of software and hardware components), and thus for increasing productivity and for providing an automated modeldriven design-flow. This paper describes recent work developed in the context of MATISSE, a MATLAB to C compiler targeting embedded systems. We introduce several techniques to allow the efficient generation of C code, such as weak types, primitives and matrix views. We evaluate the compiler with a set of 9 publicly available benchmarks, targeting both embedded systems and a desktop system. We compare the execution time of the generated C code with the original code running on MATLAB, achieving a geometric mean speedup of 8.1 ×, and qualitatively compare our results with the performance of related approaches. The use of the new techniques allowed the compiler to achieve performance improvements of 46% on average.

Abstract
In this busy society of ours people push their limits to work better and more in order to remain competitive with their peers. Nonetheless, working longer hours does not necessarily improves productivity nor performance. In order to prevent the negative consequences of this increasing trend, the evolution of performance throughout the day of work should be more closely monitored. This could avoid undesirable states or even breakdowns, which have social and economical implications. In this work we measure user performance through their interaction with the computer. We monitor its evolution during a day of work and how different types of music may increase or decrease its natural daily degradation. We conclude that the relationship between types of music and its effects is not universal and depends, among other things, on the musical profile of the individual. A prototype for a distributed music recommendation service is presented that suggests musics at an individual and group level, based on user musical profiles and objectives.

Abstract
As a recently increasing trend among different applications of smart grid vision, smart households as a new implementation area of demand response (DR) strategies have drawn more attention both in research and in engineering practice. On the other hand, optimum sizing of renewable energy based small scale hybrid systems is also a topic that is widely covered by the existing literature. In this study, the sizing of additional distributed generation and energy storage systems to be applied in smart households, which due to DR activities have a different daily demand profile compared with normal household profiles, is investigated. To the best knowledge of the authors, this is the first attempt in the literature to consider the impact of DR on sizing. The study is conducted using a mixed-integer linear programming framework for home energy management system modeling and techno-economical sizing. Also, different sensitivity analyses considering the impacts of variation of economic inputs on the provided model are realized.

Abstract
The work reported in this paper introduces a periodic switching technique applied to continuous-time filters, whose outcome is an equivalent filter with scaled time-constants. The principle behind the method is based on a procedure that extends the integration time by periodically interrupting the normal integration of the filter. The net result is an up scaling of the time constant, inversely proportional to the switching duty-cycle. This is particularly suitable for reducing the area occupied by passive devices in integrated circuits, as well as to accurately calibrate the filter dynamics. Previous works have been following this concept in an entirely continuous-time perspective, either focusing on specific circuits or using approximations to provide an extended analysis. This paper includes input/output sampling to derive a closed-form representation for the scaling technique herein coined as 'Filter & Hold' (F&H). A detailed mathematical analysis is described, demonstrating that the F&H concept represents an exact filtering solution. Simulation results and experimental measurements are provided to further validate the theoretical analysis for an F&H vector-filter prototype. Copyright (C) 2014 John Wiley & Sons, Ltd.