Abstract

Abstract
The problem of analyzing massive graph streams in real time is growing along with the size of streams. Sampling techniques have been used to analyze these streams in real time. However, it is difficult to answer questions like, which structures are well preserved by the sampling techniques over the evolution of streams? Which sampling techniques yield proper estimates for directed and weighted graphs? Which techniques have least time complexity etc? In this work, we have answered the above questions by comparing and analyzing the evolutionary samples of such graph streams. We have evaluated sequential sampling techniques by comparing the structural metrics from their samples. We have also presented a biased version of reservoir sampling, which shows better comparative results in our scenario. We have carried out rigorous experiments over a massive stream of 3 hundred million calls made by 11 million anonymous subscribers over 31 days. We evaluated node based and edge based methods of sampling. We have compared the samples generated by using sequential algorithms like, space saving algorithm for finding topK items, reservoir sampling, and a biased version of reservoir sampling. Our overall results and observations show that edge based samples perform well in our scenario. We have also compared the distribution of degrees and biases of evolutionary samples.

Abstract
Mobile phones are powerful tools to connect people. The streams of Call Detail Records (CDR’s) generating from these devices provide a powerful abstraction of social interactions between individuals, representing social structures. Call graphs can be deduced from these CDRs, where nodes represent subscribers and edges represent the phone calls made. These graphs may easily reach millions of nodes and billions of edges. Besides being large-scale and generated in real-time, the underlying social networks are inherently complex and, thus, difficult to analyze. Conventional data analysis performed by telecom operators is slow, done by request and implies heavy costs in data warehouses. In face of these challenges, real-time streaming analysis becomes an ever increasing need to mobile operators, since it enables them to quickly detect important network events and optimize business operations. Sampling, together with visualization techniques, are required for online exploratory data analysis and event detection in such networks. In this chapter, we report the burgeoning body of research in network sampling, visualization of streaming social networks, stream analysis and the solutions proposed so far. © 2016, Springer International Publishing Switzerland.

Abstract
Portable digital devices equipped with GPS antennas are ubiquitous sources of continuous information for location-based Expert and Intelligent Systems. The availability of these traces on the human mobility patterns is growing explosively. To mine this data is a fascinating challenge which can produce a big impact on both travelers and transit agencies. This paper proposes a novel incremental framework to maintain statistics on the urban mobility dynamics over a time-evolving origin-destination (O-D) matrix. The main motivation behind such task is to be able to learn from the location-based samples which are continuously being produced, independently on their source, dimensionality or (high) communicational rate. By doing so, the authors aimed to obtain a generalist framework capable of summarizing relevant context-aware information which is able to follow, as close as possible, the stochastic dynamics on the human mobility behavior. Its potential impact ranges Expert Systems for decision support across multiple industries, from demand estimation for public transportation planning till travel time prediction for intelligent routing systems, among others. The proposed methodology settles on three steps: (i) Half-Space trees are used to divide the city area into dense subregions of equal mass. The uncovered regions form an O-D matrix which can be updated by transforming the trees'leaves into conditional nodes (and vice-versa). The (ii) Partioning Incremental Algorithm is then employed to discretize the target variable's historical values on each matrix cell. Finally, a (iii) dimensional hierarchy is defined to discretize the domains of the independent variables depending on the cell's samples. A Taxi Network running on a mid-sized city in Portugal was selected as a case study. The Travel Time Estimation (TTE) problem was regarded as a real-world application. Experiments using one million data samples were conducted to validate the methodology. The results obtained highlight the straightforward contribution of this method: it is capable of resisting to the drift while still approximating context-aware solutions through a multidimensional discretization of the feature space. It is a step ahead in estimating the real-time mobility dynamics, regardless of its application field.

Abstract
In bioacoustic recognition approaches, a "flat" classifier is usually trained to recognize several species of anuran, where the number of classes is equal to the number of species. Consequently, the complexity of the classification function increases proportionally to the amount of species. To avoid this issue we propose a "hierarchical" approach that decomposes the problem into three taxonomic levels: the family, the genus, and the species level. To accomplish this, we transform the original single-label problem into a multi-dimensional problem (multi-label and multi-class) considering the Linnaeus taxonomy. Then, we develop a top-down method using a set of classifiers organized as a hierarchical tree. Thus, it is possible to predict the same set of species as a flat classifier, and additionally obtain new information about the samples and their taxonomic relationship. This helps us to understand the problem better and achieve additional conclusions by the inspection of the confusion matrices at the three levels of classification. In addition, we carry out our experiments using a Cross-Validation performed by individuals. This form of CV avoids mixing syllables that belong to the same specimens in the testing and training sets, preventing an overestimate of the accuracy and generalizing the predictive capabilities of the system. We tested our system in a dataset with sixty individual frogs, from ten different species, eight genus, and four families, achieving a final Micro-and Average-accuracy equal to 86% and 62% respectively.

Abstract
Decision rules are one of the most expressive and interpretable models for machine learning. In this article, we present Adaptive Model Rules (AMRules), the first stream rule learning algorithm for regression problems. In AMRules, the antecedent of a rule is a conjunction of conditions on the attribute values, and the consequent is a linear combination of the attributes. In order to maintain a regression model compatible with the most recent state of the process generating data, each rule uses a Page-Hinkley test to detect changes in this process and react to changes by pruning the rule set. Online learning might be strongly affected by outliers. AMRules is also equipped with outliers detection mechanisms to avoid model adaption using anomalous examples. In the experimental section, we report the results of AMRules on benchmark regression problems, and compare the performance of our system with other streaming regression algorithms.