Alan Valejo
University of São Paulo
Network
Latest external collaboration on country level. Dive into details by clicking on the dots.
Publication
Featured researches published by Alan Valejo.
acm symposium on applied computing | 2015
Jorge Carlos Valverde-Rebaza; Alan Valejo; Lilian Berton; Thiago de Paulo Faleiros; Alneu de Andrade Lopes
Link prediction in online social networks is useful in numerous applications, mainly for recommendation. Recently, different approaches have considered friendship groups information for increasing the link prediction accuracy. Nevertheless, these approaches do not consider the different roles that common neighbors may play in the different overlapping groups that they belong to. In this paper, we propose a new approach that uses overlapping groups structural information for building a naïve Bayes model. From this proposal, we show three different measures derived from the common neighbors. We perform experiments for both unsupervised and supervised link prediction strategies considering the link imbalance problem. We compare sixteen measures in four well-known online social networks: Flickr, LiveJournal, Orkut and Youtube. Results show that our proposals help to improve the link prediction accuracy.
international database engineering and applications symposium | 2014
Alan Valejo; Jorge Carlos Valverde-Rebaza; Brett Drury; Alneu de Andrade Lopes
The multilevel graph partitioning strategy aims to reduce the computational cost of the partitioning algorithm by applying it on a coarsened version of the original graph. This strategy is very useful when large-scale networks are analyzed. To improve the multilevel solution, refinement algorithms have been used in the uncorsening phase. Typical refinement algorithms exploit network properties, for example minimum cut or modularity, but they do not exploit features from domain specific networks. For instance, in social networks partitions with high clustering coefficient or similarity between vertices indicate a better solution. In this paper, we propose a refinement algorithm (RSim) which is based on neighborhood similarity. We compare RSim with: 1. two algorithms from the literature and 2. one baseline strategy, on twelve real networks. Results indicate that RSim is competitive with methods evaluated for general domains, but for social networks it surpasses the competing refinement algorithms.
brazilian conference on intelligent systems | 2014
Alan Valejo; Jorge Carlos Valverde-Rebaza; Alneu de Andrade Lopes
Many real world complex networks have an a overlapping community structure, in which a vertex belongs to one or more communities. Numerous approaches for crisp overlapping community detection were proposed in the literature, most of them have a good accuracy but their computational costs are considerably high and infeasible for large-scale networks. Since the multilevel approach has not been previously applied to deal with overlapping communities detection problem, in this paper we propose an adaptation of this approach to tackle the detection problem to overlapping communities case. The goal is to analyze the time impact and the quality of solution of our multilevel strategy regarding to traditional algorithms. Our experiments show that our proposal consistently produces good performance compared to single-level algorithms and in less time.
Neurocomputing | 2017
Lilian Berton; Thiago de Paulo Faleiros; Alan Valejo; Jorge Carlos Valverde-Rebaza; Alneu de Andrade Lopes
Abstract Graph-based semi-supervised learning (SSL) provides a powerful framework for the modeling of manifold structures in high-dimensional spaces. Additionally, graph representation is effective for the propagation of the few initial labels existing in training data. Graph-based SSL requires robust graphs as input for an accurate data mining task, such as classification. In contrast to most graph construction methods, which ignore the labeled instances available in SSL scenarios, a previous study proposed a graph-construction method, named GBILI, to exploit the informativeness conveyed by such instances available in a semi-supervised classification domain. Here, we have improved the method proposing an optimized algorithm referred to as Robust Graph that Considers Labeled Instances (RGCLI) for the generation of more robust graphs. The contributions of this paper are threefold: i) reduction of GBILI time complexity from quadratic to O ( nk log n ) . This enhancement allows addressing large datasets; ii) demonstration of RGCLI mathematical properties, proving the constructed graph is an optimal graph to model the smoothness assumption of SSL; and iii) evaluation of the efficacy of the proposed approach in a comprehensive semi-supervised classification scenario with several datasets, including an image segmentation task, which needs a large graph to represent the image. Such experiments show the use of labeled vertices in the graph construction process improves the graph topology, hence, the learning task in which it will be employed.
processing of the portuguese language | 2014
Alan Valejo; Brett Drury; Jorge Carlos Valverde-Rebaza; Alneu de Andrade Lopes
The grouping of related verbs is a mature problem in linguistics and natural language processing. There have been a number of resources which have grouped together English verbs, for example VerbNet. In comparison Portuguese has fewer resources, some of which have been based upon English verb studies. The manual grouping of Portuguese verbs would be a manually intensive task, consequently this paper presents a strategy for grouping Portuguese verbs. The strategy connects verbs through common arguments and uses overlapping community detection algorithm to identify related verbs.
Applied Intelligence | 2017
Vinícius Pereira Gonçalves; Eduardo P. Costa; Alan Valejo; Geraldo P. R. Filho; Thienne Johnson; Gustavo Pessin; Jo Ueyama
Computer systems are a part of everyday life, since they influence human behavior and stimulate changes in the emotional states of the users. The assessment of users’ emotions during their interaction with computer systems can help to provide tailorable website interfaces and better recommendations systems. However, emotions are complex and difficult to identify or assess. Previous studies have shown that, in a real-world scenario, the use of single sensors do not provide an accurate emotional assessment. Hence, in this study, we propose a framework that takes into account multiple sensors so that conclusions can be drawn about the emotional state of the user at the time of interaction. The proposed multi-sensing approach includes several inputs from users (such as speech, facial movements, and everyday activities), and uses an artificial intelligent strategy to map these different responses into one or more emotional states. The Componential Emotion Theory and Scherer’s Emotional Semantic Space are used to underpin the theoretical framework. The experimental results show that the combination of outputs generated by multiple sensors provides a more accurate assessment of emotional states than when the sensors are treated individually.
processing of the portuguese language | 2014
Brett Drury; Paula Christina Figueira Cardoso; Jorge Carlos Valverde-Rebaza; Alan Valejo; Fabio R. Pereira; Alneu de Andrade Lopes
Manually annotated data is the basis for a large number of tasks in natural language processing as either: evaluation or training data. The annotation of large amounts of data by dedicated full-time annotators can be an expensive task, which may be beyond the budgets of many research projects. An alternative is crowd-sourcing where annotations are split among many part time annotators. This paper presents a freely available open-source platform for crowd-sourcing manual annotation tasks, and describes its application to annotating causative relations.
Knowledge Based Systems | 2018
Alan Valejo; Maria Cristina Ferreira de Oliveira; Geraldo P. R. Filho; Alneu de Andrade Lopes
Abstract Multilevel approaches aim at reducing the cost of a target algorithm over a given network by applying it to a coarsened (or reduced) version of the original network. They have been successfully employed in a variety of problems, most notably community detection. However, current solutions are not directly applicable to bipartite networks and the literature lacks studies that illustrate their application for solving multilevel optimization problems in such networks. This article addresses this gap and introduces a multilevel optimization approach for bipartite networks and the implementation of a general multilevel framework including novel algorithms for coarsening and uncorsening, applicable to a variety of problems. We analyze how the proposed multilevel strategy affects the topological features of bipartite networks and show that a controlled coarsening strategy can preserve properties such as degree and clustering coefficient centralities. The applicability of the general framework is illustrated in two optimization problems, one for solving the Barber’s modularity for community detection and the second for dimensionality reduction in text classification. We show that the solutions thus obtained are statistically equivalent, regarding accuracy, to those of conventional approaches, whilst requiring considerably lower execution times.
Computer Networks | 2018
Geraldo P. R. Filho; Leandro A. Villas; Heitor Freitas; Alan Valejo; Daniel L. Guidoni; Jo Ueyama
Abstract This article proposes ResiDI, an intelligent decision-making system for a residential distributed automation infrastructure based on wireless sensors and actuators. ResiDI transmits events using wireless technologies embedded in WSANs to reduce the wire load capacity of traditional systems. In addition, the nodes are equipped with batteries, as a backup system. These features allow the ResiDI to be installed anywhere in the house, without the need for drilling or changing any other pre-existing infrastructure. Furthermore, the roles and intelligence of ResiDI are distributed among the network nodes. Besides increasing precision in decision-making through a neural network, the ResiDI seeks to reduce node energy consumption by means of a temporal correlation mechanism. As proof of concept, a prototype was developed to integrate with ResiDI in order to demonstrate its viability. When compared with an approach in the literature, real and simulated results show that ResiDI makes three key contributions: (i) 22.03% increase in decision-making; (ii) 44.35% reduction in node energy consumption in a homogeneous way; and (iii) 95.24% efficiency in information transmission. Finally, ResiDI provides a gain in response time of 30.21%, so that the decision-making process is performed faster.
Annual International Symposium on Information Management and Big Data | 2017
Alan Valejo; Vinícius Leati de Rossi Ferreira; Maria Cristina Ferreira de Oliveira; Alneu de Andrade Lopes
Interest in algorithms for community detection in networked systems has increased over the last decade, mostly motivated by a search for scalable solutions capable of handling large-scale networks. Multilevel approaches provide a potential solution to scalability, as they reduce the cost of a community detection algorithm by applying it to a coarsened version of the original network. The solution obtained in the small-scale network is then projected back to the original large-scale model to obtain the desired solution. However, standard multilevel methods are not directly applicable to bipartite networks and there is a gap in existing literature on multilevel optimization applied to such networks. This article addresses this gap and introduces a novel multilevel method based on one-mode projection that allows executing traditional multilevel methods in bipartite network models. The approach has been validated with an algorithm for community detection that solves the Barber’s modularity problem. We show it can scale a target algorithm to handling larger networks, whilst preserving solution accuracy.