Abstract
In classification problems, a dataset is said to be imbalanced when the distribution of the target variable is very unequal. Classes contribute unequally to the decision boundary, and special metrics are used to evaluate these datasets. In previous work, we presented pairwise ranking as a method for binary imbalanced classification, and extended to the ordinal case using weights. In this work, we extend ordinal classification using traditional balancing methods. A comparison is made against traditional and ordinal SVMs, in which the ranking adaption proposed is found to be competitive.

Abstract
The increasing demand for human activity analysis on surveillance scenarios has been provoking the emerging of new features and concepts that could help to identify the activities of interest. In this paper, we present a context-based descriptor to identify individual profiles. It accounts with a multi-scale histogram representation of position-based and attention-based features that follow a key-point trajectory sampling. The notion of profile is expressed by a new semantic concept introduced as an adjective for action recognition. We also identify a very rich dataset, in terms of intensity and variability of human activity, and extended it by manual annotation to validate the introduced concept of profile and test the descriptor's discriminative power. High rates of recognition were achieved.

Abstract
Optical Music Recognition (OMR) systems are an indispensable tool to transform the paper-based music scores and manuscripts into a machine-readable symbolic format. A system like this potentiates search, retrieval and analysis. One of the problematic stages is the musical symbols detection where operations to localize and to isolate musical objects are developed. The complexity is caused by printing and digitalization, as well as the paper degradation over time. Distortions inherent in staff lines, broken, connected and overlapping symbols, differences in sizes and shapes, noise, and zones of high density of symbols is even worst when we are dealing with handwritten music scores. In this paper the exploration of an optimization approach to support semantic and syntactic consistency after the music symbols extraction phase is proposed. The inclusion of this ongoing technique can lead to better results and encourage further experiences in the field of handwritten music scores recognition.

Abstract
The conservative treatment is now the preferred procedure to treat breast cancer mainly due to better aesthetical results obtained. However, the aesthetic outcome is diverse and very difficult to evaluate, which motivates the research on automatic methodologies. The use of three-dimensional (3D) methodologies is increasing; however, the high cost of the equipment and the need for specialised technicians to operate it are import setbacks. Consequently, the search for affordable and easy to perform equipments is highly desirable. This paper studies the application of a Kinect device in this field, addressing issues related to accuracy, resolution and quality of the data. The paper demonstrates a comparative study of state-of-the-art Super-Resolution (SR) algorithms applied to the Kinect depth data, and the importance to improve the quality of images is stressed. The results demonstrate that it is possible to measure volumetric information and that there is agreement between features and the subjective aesthetic evaluation.

Abstract
We present a complete and modular framework that extract trajectories in a real and complex retail scenario, under unconstrained video conditions. Two motion tracking algorithms that combine features from crowd motion detection and multiple tracking are presented to build motion patterns and understand customer's behavior. Their evaluation across several datasets show promising results.

Abstract
In 802.11 Wireless Networks, detecting faulty equipment, poor radio conditions, and changes in user behavior through anomaly detection techniques is of great importance in network management. The traffic load and user movement on different access points (APs) in a wireless covered area vary with time, making these network management tasks harder. We intend to inspect the evolving structure of wireless networks and their inherent dynamics in order to provide models for anomaly detection. For this purpose we explore the temporal usage behavior of the network by applying various types of Hidden Markov Models. We observe the usage pattern of up to 100 APs in one week period in 2011 at the Faculty of Engineering of the University of Porto. The first step of this study consists of constructing various Hidden Markov Models from 802.11 AP usage data. We then apply statistical techniques for outlier detection and justify the presented outliers by inspecting the models' parameters and a set of HMM indicators. We finally introduce examples of wireless networks anomalous patterns based on the transitions between HMM states and provide an analysis of the entire set of APs under study.