Artículos - Ciencias de la Computación
Permanent URI for this collection
Browse
Recent Submissions
Item Unsupervised detection of disturbances in 2D radiographs(IEEE Computer Society, 2021) Estacio, Laura; Mora, Rensso; Moritz, Ehlke; Lamecker, Hans; Tack, Alexander; Zachow, Stefan; Castro, Eveling"We present a method based on a generative model for detection of disturbances such as prosthesis, screws, zippers, and metals in 2D radiographs. The generative model is trained in an unsupervised fashion using clinical radiographs as well as simulated data, none of which contain disturbances. Our approach employs a latent space consistency loss which has the benefit of identifying similarities, and is enforced to reconstruct X-rays without disturbances. In order to detect images with disturbances, an anomaly score is computed also employing the Frechet distance between the input X-ray and the reconstructed one using our generative model. Validation was performed using clinical pelvis radiographs. We achieved an AUC of 0.77 and 0.83 with clinical and synthetic data, respectively. The results demonstrated a good accuracy of our method for detecting outliers as well as the advantage of utilizing synthetic data. © 2021 IEEE"Item ICE: A visual analytic tool for interactive clustering ensembles generation(Association for Computing Machinery, 2021) Castro-Ochante, J.; Camara-Chavez, G.; Gomez-Nieto, E."Clustering methods are the most used algorithms for unsupervised learning. However, there is no unique optimal approach for all datasets since different clustering algorithms produce different partitions. To overcome this issue of selecting an appropriate technique and its corresponding parameters, cluster ensemble strategies are used for improving accuracy and robustness by a weighted combination of two or more approaches. However, this process is often carried out almost in a blind manner, testing different combinations of methods and assessing if its performance is beneficial for the defined purpose. Thus, the procedure for selecting the best combination tests many clustering ensembles until the desired result is achieved. This paper proposes a novel analytic tool for clustering ensemble generation, based on quantitative metrics and interactive visual resources. Our approach allows the analysts to display different results from state-of-the-art clustering methods and analyze their performance based on specific metrics and visual inspection. Based on their requirements/experience, the analysts can interactively assign weights to the different methods to set their contributions and manage (create, store, compare, and merge), such as for ensembles. Our approach's effectiveness is shown through a set of experiments and case studies, attesting to its usefulness in practical applications. © 2021 ACM."Item Political discourses, ideologies, and online coalitions in the Brazilian Congress on Twitter during 2019(SAGE Publications Ltd, 2021) García-Sánchez, E.; Benetti, P.R.; Higa, G.L.; Alvarez, M.C.; Gomez-Nieto, E."The aim of this research is to describe the pattern of interactions of Brazilian legislators on Twitter during 2019 in the construction of political discourses. Based on 20,076 replies during 2019, posted on Twitter by 514 Brazilian legislators, we conducted descriptive analysis of legislators’ Twitter profiles, social network analyses from their interactions, and content analysis of the messages. We found that (1) there are large disparities between legislators in the use of Twitter; (2) the pattern of interactions depicted five clusters defined by political affinities; (3) each cluster had different features regarding their composition and impact; (4) the centrality of the legislators within the network was positively associated with public endorsement on Twitter; and (5) the topics of messages within the clusters reinforce discourses aligned to political ideologies. We argue that the pattern of interactions on Twitter allows to identify online coalitions that reinforce particular discourses within the Brazilian parliamen"Item AutoImplant 2020-First MICCAI Challenge on Automatic Cranial Implant Design(Institute of Electrical and Electronics Engineers Inc., 2021) Li, Jianning; Pimentel, Pedro; Szengel, Angelika; Ehlke, Moritz; Lamecker, Hans; Zachow, Stefan; Estacio, Laura; Doenitz, Christian; Ramm, Heiko; Shi, Haochen; Chen, Xiaojun; Matzkin, Franco; Newcombe, Virginia; Ferrante, Enzo; Jin, Yuan; Ellis, David G.; Aizenberg, Michele R.; Kodym, Oldrich; Spanel, Michal; Herout, Adam; Mainprize, James G; Fishman, Zachary; Hardisty, Michael R.; Bayat, Amirhossein; Shit, Suprosanna; Wang, Bomin; Liu, Zhi; Eder, Matthias; Pepe, Antonio; Gsaxner, Christina; Alves, Victor; Zefferer, Ulrike; von Campe, Gord; Pistracher, Karin; Schafer, Ute; Schmalstieg, Dieter; Menze, Bjoern H.; Glocker, Ben; Egger, JanThe aim of this paper is to provide a comprehensive overview of the MICCAI 2020 AutoImplant Challenge. The approaches and publications submitted and accepted within the challenge will be summarized and reported, highlighting common algorithmic trends and algorithmic diversity. Furthermore, the evaluation results will be presented, compared and discussed in regard to the challenge aim: seeking for low cost, fast and fully automated solutions for cranial implant design. Based on feedback from collaborating neurosurgeons, this paper concludes by stating open issues and post-challenge requirements for intra-operative use. The codes can be found at https://github.com/Jianningli/tmi. © 1982-2012 IEEEItem A Comparison of Machine Learning Classifiers for Water-Body Segmentation Task in the PeruSAT-1 Imagery(Springer Science and Business Media Deutschland GmbH, 2021) Huauya, R.; Moreno, F.; Peña, J.; Dianderas, E.; Mauricio, A.; Díaz, J,"Water-body segmentation is a high-relevance task inside satellite image analysis due to its relationship with environmental monitoring and assessment. Thereon, several authors have proposed different approaches which achieve a wide range of results depending on their datasets and settings. This study is a brief review of classical segmentation techniques in multispectral images using the Peruvian satellite PeruSAT-1 imagery. The areas of interest are medium-sized highland zones with water bodies around in Peruvian south. We aim to analyze classical segmentation methods to prevent future natural disasters, like alluviums or droughts, under low-cost data constraints. We consider accuracy, robustness, conditions, and visual effects in our analysis"Item CrimAnalyzer: Understanding Crime Patterns in Sao Paulo(IEEE COMPUTER SOC10662 LOS VAQUEROS CIRCLE, PO BOX 3014, LOS ALAMITOS, CA 90720-1314, 2021) Garcia, Germain; Silveira, Jaqueline; Poco, Jorge; Paiva, Afonso; Nery, Marcelo Batista; Silva, Claudio T.; Adorno, Sergio; Nonato, Luis GustavoSao Paulo is the largest city in South America, with crime rates that reflect its size. The number and type of crimes vary considerably around the city, assuming different patterns depending on urban and social characteristics of each particular location. Previous works have mostly focused on the analysis of crimes with the intent of uncovering patterns associated to social factors, seasonality, and urban routine activities. Therefore, those studies and tools are more global in the sense that they are not designed to investigate specific regions of the city such as particular neighborhoods, avenues, or public areas. Tools able to explore specific locations of the city are essential for domain experts to accomplish their analysis in a bottom-up fashion, revealing how urban features related to mobility, passersby behavior, and presence of public infrastructures (e.g., terminals of public transportation and schools) can influence the quantity and type of crimes. In this paper, we present CrimAnalyzer, a visual analytic tool that allows users to study the behavior of crimes in specific regions of a city. The system allows users to identify local hotspots and the pattern of crimes associated to them, while still showing how hotspots and corresponding crime patterns change over time. CrimAnalyzer has been developed from the needs of a team of experts in criminology and deals with three major challenges: i) flexibility to explore local regions and understand their crime patterns, ii) identification of spatial crime hotspots that might not be the most prevalent ones in terms of the number of crimes but that are important enough to be investigated, and iii) understand the dynamic of crime patterns over time. The effectiveness and usefulness of the proposed system are demonstrated by qualitative and quantitative comparisons as well as by case studies run by domain experts involving real data. The experiments show the capability of CrimAnalyzer in identifying crime-related phenomena.Item Urban Perception: Can We Understand Why a Street Is Safe?(Springer Science and Business Media Deutschland GmbH, 2021) Moreno-Vera, Felipe; Lavi, Bahram; Poco, JorgeThe importance of urban perception computing is relatively growing in machine learning, particularly in related areas to Urban Planning and Urban Computing. This field of study focuses on developing systems to analyze and map discriminant characteristics that might direcly impact the city’s perception. In other words, it seeks to identify and extract discriminant components to define the behavior of a city’s perception. This work will perform a street-level analysis to understand safety perception based on the “visual components”. As our result, we present our experimental evaluation regarding the influence and impact of those visual components on the safety criteria and further discuss how to properly choose confidence on safe or unsafe measures concerning the perceptional scores on the city street levels analysis. © 2021, Springer Nature Switzerland AGItem Understanding safety based on urban perception(Springer Science and Business Media Deutschland GmbH, 2021) Moreno-Vera, F."Currently, one important field on machine learning is Urban Perception Computing is to model the way in which humans can interact and understand the environment that surrounds them. This process is performed using convolutional models to learn and identify some insights which define the concept of perception of a place (e.g. a street image). One approach of this field is urban perception of street images, we will focus on this approach to study the safety perception of a city and try to explain why and how the perception can be predicted by a mathematical model. As result, we present an analysis about the influence and impact of the visual components on the safety criteria and also an explanation about why a certain decision on the perception of the safety of the streets, such as safe or unsafe. © 2021, Springer Nature Switzerland AG"Item Ontoslam: An ontology for representing location and simultaneous mapping information for autonomous robots(MDPI, 2021) Cornejo-Lupa, M.A.; Cardinale, Y.; Ticona-Herrera, R.; Barrios Aranibar, D.; Andrade, M.; Diaz-Amado, J."Autonomous robots are playing an important role to solve the Simultaneous Localization and Mapping (SLAM) problem in different domains. To generate flexible, intelligent, and interopera-ble solutions for SLAM, it is a must to model the complex knowledge managed in these scenarios (i.e., robots characteristics and capabilities, maps information, locations of robots and landmarks, etc.) with a standard and formal representation. Some studies have proposed ontologies as the standard representation of such knowledge; however, most of them only cover partial aspects of the information managed by SLAM solutions. In this context, the main contribution of this work is a complete ontology, called OntoSLAM, to model all aspects related to autonomous robots and the SLAM problem, towards the standardization needed in robotics, which is not reached until now with the existing SLAM ontologies. A comparative evaluation of OntoSLAM with state-of-the-art SLAM ontologies is performed, to show how OntoSLAM covers the gaps of the existing SLAM knowledge representation models. Results show the superiority of OntoSLAM at the Domain Knowledge level and similarities with other ontologies at Lexical and Structural levels. Additionally, OntoSLAM is integrated into the Robot Operating System (ROS) and Gazebo simulator to test it with Pepper robots and demonstrate its suitability, applicability, and flexibility. Experiments show how OntoSLAM provides semantic benefits to autonomous robots, such as the capability of inferring data from organized knowledge representation, without compromising the information for the application and becoming closer to the standardization needed in robotics. © 2021 by the authors. Licensee MDPI, Basel, Switzerland"Item A multi-modal visual emotion recognition method to instantiate an ontology(SciTePress, 2021) A. Heredia, Juan Pablo; Cardinale, Yudith; Dongo, Irvin; Díaz-Amado, Jose"Human emotion recognition from visual expressions is an important research area in computer vision and machine learning owing to its significant scientific and commercial potential. Since visual expressions can be captured from different modalities (e.g., face expressions, body posture, hands pose), multi-modal methods are becoming popular for analyzing human reactions. In contexts in which human emotion detection is performed to associate emotions to certain events or objects to support decision making or for further analysis, it is useful to keep this information in semantic repositories, which offers a wide range of possibilities for implementing smart applications. We propose a multi-modal method for human emotion recognition and an ontology-based approach to store the classification results in EMONTO, an extensible ontology to model emotions. The multi-modal method analyzes facial expressions, body gestures, and features from the body and the environment to determine an emotional state; this processes each modality with a specialized deep learning model and applying a fusion method. Our fusion method, called EmbraceNet+, consists of a branched architecture that integrates the EmbraceNet fusion method with other ones. We experimentally evaluate our multi-modal method on an adaptationof the EMOTIC dataset. Results show that our method outperforms the single-modal methods."Item DBM*-Tree: An efficient metric acces method(Scopus, 2007) Ocsa, Alexander; Cuadros Vargas, ErnestoIn this paper we propose a new dynamic Metric Access Method (MAM) called DBM*-Tree, which uses precomputed distances to reduce the construction cost avoiding repeated calculus of distance. Making use of the pre-calculated distances cost of similarity queries are also reduced by taking various local representative objects in order to increment the pruning of irrelevant elements during the query. We also propose a new algorithm to select the suitable subtree in the insertion operation, which is an evolution of the previous methods. Empiric tests on real and synthetic data have shown evidence that DBM*-Tree requires 25 % less average distance computing than Density Based Metric Tree (DBM-Tree) which is one of the most efficient and recent MAM found in the literature. © Copyright 2007 ACM.Item DB-GNG: A constructive self-organizing map based on density(Scopus, 2007) Ocsa, Alexander; Bedregal, Carlos; Cuadros Vargas, ErnestoNowadays applications require efficient and fast techniques due to the growing volume of data and its increasing complexity. Recent studies promote the use of Access Methods (AMs) with Self-Organizing Maps (SOMs) for a faster similarity information retrieval. This paper proposes a new constructive SOM based on density, which is also useful for clustering. Our algorithm creates new units based on density of data, producing a better representation of the data space with a less computational cost for a comparable accuracy. It also uses AMs to reduce considerably the Number of Distance Calculations during the training process, outperforming existing constructive SOMs by as much as 89%. ©2007 IEEE.Item Improving human computer interaction through spoken natural language(Scopus, 2007) Florez Choque, Omar; Cuadros Vargas, ErnestoThe fastest and most straightforward way of communication for mankind is the voice. Therefore, the best way to interact with computers should be the voice too. That is why at the moment men are searching new ways to interact with computers. This interaction is improved if the words spoken by the speaker are organized in Natural Language. In this article, it is proposed a model to recover information from databases through queries in Spanish Natural Language using the voice as the way of communication. This model incorporates a Hybrid Intelligent System based on Genetic Algorithms and a Kohonen Self-Organizing Map (SOM) to recognize the present phonemes in a word through time. This approach allows us to remake up a word with speaker independence. Furthermore, it is proposed the use of a compiler with type 2 grammar according to the Chomsky Hierarchy to support the syntactic and semantic structure in Spanish language. Our experiments suggest that the Spoken Natural Language improves notably the Human-Computer interaction when compared with traditional input methods such as: mouse or keybord. © 2007 IEEE.Item Controlling oil production in smart wells by MPC strategy with reinforcement learning(Scopus, 2010) Talavera, Alvaro; Túpac Valdivia, Yván Jesús; Vellasco, MarleyThis work presents the modeling and development of a methodology based on Model Predictive Control - MPC that uses a machine learning model, based on Reinforcement Learning, as the method for searching the optimal control policy, and a neural network as a proxy, for modeling the nonlinear plant. The neural network model was developed to predict the following variables: average pressure of the reservoir, the daily production of oil, gas, water and water cut in the production well, for three consecutive values, to perform the predictive control. This model is applied as a strategy to control the oil production in an oil reservoir with existing producer and injector wells. The experiments were carried out on a synthetic oil reservoir model that consists in a reservoir with three layers with different permeability and one producer well and one injector well, both completed in the three layers. There are three valves located into the injector well, one for each completion, which are the handling variables of the model. The oil production of the producer well is the controlled variable. The experiments performed have considered various set points and also the impact of disturbances on the production well. The obtained results indicate that the proposed model is capable of controlling oil production even with disturbances in the producing well, for different reference values for oil production and supporting some features of the petroleum reservoir systems such as: strong non-linearity, long delay in the system response, and multivariate characteristic. Copyright 2010, Society of Petroleum Engineers.Item Learning how to extract rotation-invariant and scale-invariant features from texture images(Scopus, 2008) Montoya Zegarra, Javier; Paulo Papa, Joao; Leite, Neucimar; da Silva Torres, Ricardo; Falcao, AlexandreLearning how to extract texture features from noncontrolled environments characterized by distorted images is a still-open task. By using a new rotation-invariant and scale-invariant image descriptor based on steerable pyramid decomposition, and a novel multiclass recognition method based on optimum-path forest, a new texture recognition system is proposed. By combining the discriminating power of our image descriptor and classifier, our system uses small-size feature vectors to characterize texture images without compromising overall classification rates. State-of-the-art recognition results are further presented on the Brodatz data set. High classification rates demonstrate the superiority of the proposed system.Item Combining global with local texture information for image retrieval applications(Scopus, 2008) Montoya Zegarra, Javier; Beeck, Jan; Jerônimo Leite, Neucimar; da Silva Torres, Ricardo; Falcao, AlexandreThis paper proposes a new texture descriptor to guide the search and retrieval in image databases. It extracts rich information from global and local primitives of textured images. At a higher level, the global macro-features in textured images are characterized by exploiting the multi-resolution properties of the Steerable Pyramid Decomposition. By doing this, the global texture configurations are highlighted. At afiner level, the local arrangements of texture micro-patterns are encoded by the Local Binary Pattern operator. Experiments were carried out on the standard Vistex dataset aiming to compare our desriptors against popular texture extraction methods with regard to their retrieval accuracies. The comparative evaluations allowed us to show the superior descriptive properties of our feature representation methods. © 2008 IEEE.Item Using large databases and self-organizing maps without tears(Scopus, 2006) Bedregal, Carlos; Cuadros Vargas, ErnestoNowadays the need to process lots of complex multimedia databases is more frequent. Recent investigations such as MAM-SOM* and SAM-SOM* families propose the combination of Self-Organizing Maps (SOM) with Access Methods for a faster similarity information retrieval. In this investigation we present experimental results using recent Access Methods such as Slim-Tree and Omni-Sequential that show the improvement acquired by these techniques and their properties in contrast with a traditional SOM network, observing up to 90% of performance improvement. © 2006 IEEE.Item Wavelet-based fingerprint image retrieval(Scopus, 2009) Montoya Zegarra, Javier; Leite, Neucimar; da Silva Torres, RicardoThis paper presents a novel approach for personal identification based on a wavelet-based fingerprint retrieval system which encompasses three image retrieval tasks, namely, feature extraction, similarity measurement, and feature indexing. We propose the use of different types of Wavelets for representing and describing the textural information presented in fingerprint images in a compact way. For that purpose, the feature vectors used to characterize the fingerprints are obtained by computing the mean and the standard deviation of the decomposed images in the wavelet domain. These feature vectors are used both to retrieve the most similar fingerprints, given a query image, and their indexation is used to reduce the search spaces of candidate images. The different types of Wavelets used in our study include: Gabor wavelets, tree-structured wavelet decomposition using both orthogonal and bi-orthogonal filter banks, as well as the steerable wavelets. To evaluate the retrieval accuracy of the proposed approach, a total number of eight different data sets were considered. We also took into account different combinations of the above wavelets with six similarity measures. The results show that the Gabor wavelets combined with the Square Chord similarity measure achieves the best retrieval effectiveness. © 2008 Elsevier B.V. All rights reserved.Item A translation from RSL to CSP(Scopus, 2008) Parisaca Vargas, Abigail; Tapia Tarifa, Silvia Lizeth; George, ChrisThe Raise Specification Language (RSL) is a broad spectrum modeling language which supports a wide range of specification styles. In order to apply verification techniques based on model checking to descriptions of concurrent systems in RSL, we translate RSL specifications into the input language CSPM of the FDR model checker. FDR is a well-established model checker for the process algebra CSP. However, we need to show that the analysis performed in FDR carry over to the original RSL specifications. For this purpose, we define a syntactic and semantic translation between RSL and CSPM, and show that this translation is in fact a strong bisimulation which preserves various properties such as traces and deadlock. Finally, we have built a tool which automates the translation of RSL specifications into CSPM following this approach. © 2008 IEEE.Item Introduction to the SAM-S M* and MAM-S M* families(Scopus, 2005) Cuadros Vargas, Ernesto; Romero, FrancelinIn this paper, two new families of constructive Self-Organizing Maps (SOMs), SAM-SOM* and MAM-SOM*, are proposed. These families are specially useful for information retrieval from large databases, high-dimensional spaces and complex distance functions which usually consume a long time. They are generated by incorporating Spatial Access Method (SAM) and Metric Access Method (MAM) into SOM with the maximum insertion rate, i.e. the case when a new unit is created for each pattern presented to the network. In this specific case, the network presents interesting advantages and acquires new properties which are quite different of traditional SOM. In a constructive SOM, if new units are rarely inserted into network, the training algorithm would probably need a long time to converge. On the other hand, if new units are inserted frequently, the training algorithm would not have enough time to adapt these units to the data distribution. Besides, training time is increased because the search for the winning neuron is traditionally performed sequentially. The use of SAM and MAM combined with SOM open the possibility of training constructive SOM with as much units as existing patterns with less time and interesting advantages compared with both models: Kohonen network SOM and SAM-SOM model (SOM using SAM). Advantages and drawbacks of these new families are also discussed. These new families are useful to improve both SOM and SAM techniques.