Abstract
Peer Observation of Teaching has raised a lot of interest as a device for quality enhancement of teaching. While much research has focused on its models, implementation schemes and feedback to the observed, little attention has been paid to what the observer actually sees and can learn from the observation. A multidisciplinary peer observation of teaching program is described, and its data is used to identify the pedagogical aspects to which lecturers pay more attention to when observing classes. The discussion addresses the valuable learning opportunities for observers provided by this program, as well as its usefulness in disseminating, sharing and clarifying quality teaching practices. The need for further research concerning teacher-student relationships and students' engagement is also suggested.

Abstract
Dealing with missing data is a crucial step in the preprocessing stage of most data mining projects. Especially in healthcare contexts, addressing this issue is fundamental, since it may result in keeping or loosing critical patient information that can help physicians in their daily clinical practice. Over the years, many researchers have addressed this problem, basing their approach on the implementation of a set of imputation techniques and evaluating their performance in classification tasks. These classic approaches, however, do not consider some intrinsic data information that could be related to the performance of those algorithms, such as features’ distribution. Establishing a correspondence between data distribution and the most proper imputation method avoids the need of repeatedly testing a large set of methods, since it provides a heuristic on the best choice for each feature in the study. The goal of this work is to understand the relationship between data distribution and the performance of well-known imputation techniques, such as Mean, Decision Trees, k-Nearest Neighbours, Self-Organizing Maps and Support Vector Machines imputation. Several publicly available datasets, all complete, were selected attending to several characteristics such as number of distributions, features and instances. Missing values were artificially generated at different percentages and the imputation methods were evaluated in terms of Predictive and Distributional Accuracy. Our findings show that there is a relationship between features’ distribution and algorithms’ performance, although some factors must be taken into account, such as the number of features per distribution and the missing rate at state.

Abstract
In order to detect network motifs we need to evaluate the exceptionality of subgraphs in a given network. This is usually done by comparing subgraph frequencies on both the original and an ensemble of random networks keeping certain structural properties. The classical null model implies preserving the degree sequence. In this paper our focus is on a richer model that approximately fixes the frequency of subgraphs of size K - 1 to compute motifs of size K. We propose a method for generating random graphs under this model, and we provide algorithms for its efficient computation. We show empirical results of our proposed methodology on neurobiological networks, showcasing its efficiency and its differences when comparing to the traditional null model.

Abstract
This paper proposes a comprehensive model for the interactions of the plug-in electric vehicles (PEVs) involved parties. An aggregator with mixed resources is assumed to be the interface between the parking lot (PL) and the upstream energy and reserve markets. On the other hand, the interactions of the PEV owners and the PL are also modeled as they impose restrictions to the PL's behavior. Therefore, a bilevel problem is constructed where in the upper level the objective of the aggregator is to maximize its profit through its interactions, and in the lower level the PL maximizes its own profit limited to the preferences of PEVs. The objectives of the upper and lower levels are contradictory; hence, an equilibrium point should be found to solve the problem. In this regard, the duality theorem is employed to convert the bilevel model to a mathematical program with equilibrium constraints. The model is implemented on the IEEE 37-bus network with added distributed generations. Various cases are thoroughly investigated and conclusions are duly drawn.

Abstract
The visual impaired are a specific minority group that can benefit from specific assistive systems in order to mitigate their mobility and accessibility constrains. In the last decade, our research group has been integrating and developing assistive technologies, focused in human-computer interaction, artificial vision, assisted navigation, pervasive computing, among others. Several projects and prototypes have been developed with the main objective of improving the blind's autonomy, mobility, and quality of life. Currently the technology has reached a maturation point that allows the development of systems based on video capturing, image recognition and location referencing, which are key for providing features of artificial vision, assisted navigation and spatial perception. The miniaturization of electronics can be used to create devices such as electronic canes that equipped with sensors can provide so much more contextual information to a blind user. The adoption of these systems is dependent of an information catalogue regarding points of interest and their physical location reference. In this paper we describe the current work on assistive systems for the blind and propose a new perspective on using the base information of those systems to provide new services to the general public. By bridging the gap between the two groups, we expect to further advance the development of the current systems and contribute to their economic sustainability.

Abstract
Formal methods technologies have the potential to verify the usability and safety of user interface (UI) software design in medical devices, enabling significant reductions in use errors and consequential safety incidents with such devices. This however depends on comprehensive and verifiable safety requirements to leverage these techniques for detecting and preventing flaws in UI software that can induce use errors. This paper presents a hazard analysis method that extends Leveson’s System Theoretic Process Analysis (STPA) with a comprehensive set of causal factor categories, so as to provide developers with clear guidelines for systematic identification of use-related hazards associated with medical devices, their causes embedded in UI software design, and safety requirements for mitigating such hazards. The method is evaluated with a case study on the Gantry-2 radiation therapy system, which demonstrates that (1) as compared to standard STPA, our method allowed us to identify more UI software design issues likely to cause use-related hazards; and (2) the identified UI software design issues facilitated the definition of precise, verifiable safety requirements for UI software, which could be readily formalized in verification tools such as Prototype Verification System (PVS). © Springer International Publishing AG (outside the US) 2017.