Pasar al contenido principal

Resultados relacionados

Resultados totales (Incluyendo duplicados): 26
Encontrada(s) 1 página(s)

Repositorio Digital de la UPF
Publicación Contribución de congreso (ConferenceOutput)

From emotions to action units with hidden and semi-hidden-task learning

  • Ruiz, Adrià
  • Van de Weijer, Joost
  • Binefa i Valls, Xavier
Comunicació presentada a: IEEE International Conference on Computer Vision (ICCV 2015) celebrat a Santiago, Xile, de l'11 al 18 de desembre de 2015., Limited annotated training data is a challenging problem in Action Unit recognition. In this paper, we investigate how the use of large databases labelled according to the 6 universal facial expressions can increase the generalization ability of Action Unit classifiers. For this purpose, we propose a novel learning framework: Hidden-Task Learning. HTL aims to learn a set of Hidden-Tasks (Action Units) for which samples are not available but, in contrast, training data is easier to obtain from a set of related Visible- Tasks (Facial Expressions). To that end, HTL is able to exploit prior knowledge about the relation between Hidden and Visible-Tasks. In our case, we base this prior knowledge on empirical psychological studies providing statistical correlations between Action Units and universal facial expressions. Additionally, we extend HTL to Semi-Hidden Task Learning (SHTL) assuming that Action Unit training samples are also provided. Performing exhaustive experiments over four different datasets, we show that HTL and SHTL improve the generalization ability of AU classifiers by training them with additional facial expression data. Additionally, we show that SHTL achieves competitive performance compared with state-of-the-art Transductive Learning approaches which face the problem of limited training data by using unlabelled test samples during training., This paper is part of a project that has received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement No 645012. Adria Ruiz and Xavier Binefa would also like to acknowledge Spanish Government to provide support under grants CICYT TIN2012-39203 and FPU13/01740. Joost van de Weijer acknowledges Project TIN2013-41751 of the Spanish Ministry of Science and the Generalitat de Catalunya Project under Grant 2014-SGR-221.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Contribución de congreso (ConferenceOutput)

Developing a depth-based tracking systems for interactive playful environments with animals

  • Pons Tomás, Patricia
  • Catalá Bolós, Alejandro
  • Jaén Martínez, Francisco Javier|||0000-0002-8815-9643
[EN] Digital games for animals within Animal Computer Interaction are usually single-device oriented, however richer interactions could be delivered by considering multimodal environments and expanding the number of technological elements involved. In these playful ecosystems, animals could be either alone or accompanied by human beings, but in both cases the system should react properly to the interactions of all the players, creating more engaging and natural games. Technologically-mediated playful scenarios for animals will therefore require contextual information about the game participants, such as their location or body posture, in order to suitably adapt the system reactions. This paper presents a depth-based tracking system for cats capable of detecting their location, body posture and field of view. The proposed system could also be extended to locate and detect human gestures and track small robots, becoming a promising component in the creation of intelligent interspecies playful environments., Work supported by the Spanish Ministry of Economy and Competitiveness and funded by the EDRF-FEDER (TIN2014-60077-R). The work of Patricia Pons has been supported by a national grant from the Spanish MECD (FPU13/03831). Alejandro Catalá also received support from a VALi+d fellowship from the GVA (APOSTD/2013/013). Special thanks to our cat participants, their owners, and our feline caretakers and therapists
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Contribución de congreso (ConferenceOutput)

Fusibles estructurales para evitar la propagación de fallos en las cimbras de edificación

  • Buitrago, Manuel|||0000-0002-5561-5104
  • Calderón García, Pedro Antonio|||0000-0002-9783-9333
  • Moragues, Juan J|||0000-0001-8111-5480
  • Oliver-Cano, Marina
  • Adam, Jose M|||0000-0002-9205-8458
[ES] La construcción de edificios con estructura de hormigón armado se realiza, habitualmente, con la ayuda de puntales metálicos que se encargan de transmitir las cargas de las plantas recién hormigonadas a las inferiores. Los principales problemas que presenta el empleo de puntales son: a) la aparición de cargas superiores a las previstas en los puntales, pudiendo provocar el colapso de toda la cimbra o incluso del edificio, b) la heterogeneidad de las cargas que reciben los puntales de una misma planta cimbrada, y c) la limitada variedad de puntales de que disponen las casas comerciales. Todo ello obliga a que, en la mayor parte de las ocasiones, los puntales empleados en las obras estén claramente sobredimensionados. En este artículo se presenta el fusible estructural para el control de la carga de los puntales (también llamado limitador de carga o LL por sus siglas en inglés), que fue desarrollado en ICITECH a partir de estudios teóricos avanzados, una intensa campaña experimental y la ejecución de una prueba de concepto. El artículo se compone de: 1) una breve recopilación de los trabajos experimentales y de validación computacional llevados a cabo; y 2) el estudio realizado con simulación numérica avanzada para evaluar y cuantificar la posible mitigación del daño y el riesgo de fallo de estructuras de edificios durante su construcción. De los resultados obtenidos se puede afirmar que con el uso de LLs se consigue una mejora: de la seguridad, frente a fallos locales de la cimbra cuyas consecuencias podrían propagarse de forma desproporcionada; de la eficiencia estructural, con una mayor uniformidad en las cargas que reciben los puntales; y de los costes del sistema de apuntalamiento., Los autores quieren expresar su gratitud a todas las entidades por la financiación recibida (FPU13/02466; PRX17/00302; GV/2015/063; APOSTD/2019/101; 101000396; IJC2020- 042642-I), así como a las empresas Encofrado J. Alsina S.A. y Levantina, Ingeniería y Construcción S.L. por su invalorable cooperación.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Contribución de congreso (ConferenceOutput)

ICFHR2016 Handwritten Keyword Spotting Competition (H-KWS 2016)

  • Pratikakis, Ioannis
  • Zagoris, Konstantinos
  • Gatos, Basilis
  • Puigcerver, Joan
  • Toselli, Alejandro Héctor|||0000-0001-6955-9249
  • Vidal, Enrique|||0000-0003-4579-5196
[EN] The H-KWS 2016, organized in the context of the ICFHR 2016 conference aims at setting up an evaluation framework for benchmarking handwritten keyword spotting (KWS) examining both the Query by Example (QbE) and the Query by String (QbS) approaches. Both KWS approaches were hosted into two different tracks, which in turn were split into two distinct challenges, namely, a segmentation-based and a segmentation-free to accommodate different perspectives adopted by researchers in the KWS field. In addition, the competition aims to evaluate the submitted training-based methods under different amounts of training data. Four participants submitted at least one solution to one of the challenges, according to the capabilities and/or restrictions of their systems. The data used in the competition consisted of historical German and English documents with their own characteristics and complexities. This paper presents the details of the competition, including the data, evaluation metrics and results of the best run of each participating methods., This work was partially supported by the Spanish MEC under FPU grant FPU13/06281, by the Generalitat Valenciana under the Prometeo/2009/014 project grant ALMA-MATER, and through the EU projects: HIMANIS (JPICH programme, Spanish grant Ref. PCIN-2015-068) and READ (Horizon-2020 programme, grant Ref. 674943).
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Contribución de congreso (ConferenceOutput)

Intelligent Playful Environments for Animals

  • Pons Tomás, Patricia
  • Catalá Bolós, Alejandro
  • Jaén Martínez, Francisco Javier|||0000-0002-8815-9643
We are evolving towards an interconnected and ubiquitous world, where digital devices and interfaces progressively adapt themselves to fit our needs and ease our daily activities. Although we coexist with plenty of animal species, such as our pets, we are approaching the evolution of technology in a strictly human-centric manner. A new field in Computer Science, called Animal-Computer Interaction (ACI), aims at filling this technological gap by developing systems and interfaces specifically designed for animals. Supporting animals' natural behavior and habits with suitable technology could improve both humans and animals' wellbeing. As a consequence, this doctoral research aims to explore, design and develop animal-centered intelligent systems that focus on enhancing one of the most natural animal behaviors: play. Therefore, the main goal of this research is to expand ACI with the ability of automatically manage and adapt animals play activity in order to improve their wellbeing., Work supported by MINECO (TIN2010-20488 and TIN2014-60077-R), UPV (UPV-FE-2014-24), MECD (FPU13/03831) and GVA (APOSTD/2013/013).
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Contribución de congreso (ConferenceOutput)

Scalable Block-Tridiagonal Eigensolvers in the Context of Electronic Structure Calculations

  • Lamas Daviña, Alejandro
  • Jose E. Roman|||0000-0003-1144-6772
  • Cartoixà, Xavier
[EN] In computer-based numerical simulations, some methods to determine the electronic and optical properties of semiconductor nanostructures, require computing the energies that correspond to the interior eigenvalues of a Hamiltonian matrix. We study the case in which the Schrödinger equation is expanded into a matrix that has a block-tridiagonal structure. Additionally, the matrix can have two extra non zero blocks in the corners due to periodic boundary conditions. Given that not the whole eigenspectrum is required, we choose to use projection methods to compute the necessary set of eigenvalues. The shift-and-invert Lanczos method requires to solve a linear system at each iteration. We have developed a parallel code that improves the scalability of this step by exploiting the block structure. Results show that, to solve these specific cases, this method offers better scalability when compared to a general-purpose solver such as MUMPS., The authors would like to acknowledge the support by Agencia Estatal de Investigacion under project TIN2016-75985-P, which includes European Commission ERDF funds, and the Spanish Ministerio de Economia y Competitividad under project No. TEC2015-67462-C2-1-R (MINECO/FEDER). Also, this work has received funding from the European Union¿s Horizon 2020 research and innovation program under grant agreement No 696656, and the Department d¿Universitats, Recerca i Societat de la Informacio of the Generalitat de Catalunya under contract 2014-SGR-384. Alejandro Lamas Davi¿na was supported by the Spanish Ministry of Education, Culture and Sport through grant FPU13-06655. The authors are also grateful for the resources provided by the Spanish Supercomputing Network.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Contribución de congreso (ConferenceOutput)

Beyond the limits of digital interaction: should animals play with interactive environments?

  • Pons Tomás, Patricia
  • Catalá Bolós, Alejandro
  • Jaén Martínez, Francisco Javier|||0000-0002-8815-9643
Our digital world evolves towards ubiquitous and intuitive scenarios, filled with interconnected and transparent computing devices which ease our daily activities. We have approached this evolution of technology in a strictly human-centric manner. There are, however, plenty of species, among them our pets, which could also profit from these technological advances. A new field in Computer Science, called Animal-Computer Interaction (ACI), aims at filling this technological gap by developing systems and interfaces specifically designed for animals. This paper envisions how ACI could be extended to enhance the most natural animal behavior: play. This work explains how interactive environments could become playful scenarios where animals enjoy, learn and interact with technology, improving their wellbeing, This work is partially funded by the Spanish Ministry of Science and Innovation under the National R&D&I Program within the project CreateWorlds (TIN2010-20488). The work of Patricia Pons is supported by an FPU fellowship from the Spanish Ministry of Education, Culture and Sports (FPU13/03831). It also received support from a postdoctoral fellowship within the VALi+d Program of the Conselleria d’Educació, Cultura I Esport (Generalitat Valenciana) awarded to Alejandro Catalá (APOSTD/2013/013). We also thank the Valencian Society for the Protection of Animals and Plants (SVPAP) for their cooperation.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Contribución de congreso (ConferenceOutput)

Problema de autovalores de la ecuación de la difusión neutrónica discretizada con el Método de Volúmenes Finitos en un PWR

  • Bernal García, Álvaro
  • Jose E. Roman|||0000-0003-1144-6772
  • Miró Herrero, Rafael|||0000-0003-1012-0869
  • Verdú Martín, Gumersindo Jesús|||0000-0001-5098-080X
En este trabajo, se ha desarrollado un algoritmo para aplicar el Método de Volúmenes Finitos (MVF) a la ecuación de la difusión neutrónica, que calcula varios autovalores y se ha incorporado al algoritmo la lectura de secciones eficaces en función de variables termohidráulicas. Para su validación, se ha simulado un reactor PWR, cuyas secciones eficaces se han modelado con la metodología SIMTAB para las condiciones termohidráulicas de Begin of Cycle (BOC) y Hot Full Power (HFP). Se ha realizado un análisis de sensibilidad de la discretización de la geometría, utilizando 3 mallas de hexaedros y otras 3 de tetraedros. En cada caso, se calculan 5 autovalores y sus correspondientes potencias., Este trabajo ha sido financiado parcialmente por el Ministerio de Educación Cultura y Deporte bajo la beca FPU13/01009, el Ministerio de Ciencia e Innovación bajo los proyectos ENE2014-59442-P y ENE2012-34585, la Generalitat Valenciana bajo el proyecto PROMETEOII/2014/008, la Universitat Politècnica de València bajo el proyecto UPPTE/2012/118, y el Ministerio de Economía y Competitividad bajo el proyecto TIN2013-41049-P.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Contribución de congreso (ConferenceOutput)

The MLLP system for the 4th CHiME Challenge

  • Del-Agua, Miguel Ángel
  • Martínez-Villaronga, Adrià
  • Giménez Pastor, Adrián
  • Sanchis Navarro, José Alberto|||0000-0002-2943-0990
  • Civera Saiz, Jorge|||0000-0002-0963-0143
  • Juan, Alfons|||0000-0002-9984-4072
[EN] The MLLP CHiME-4 system is presented in this paper. It has been built using the transLectures-UPV toolkit (TLK) developed by the MLLP research group which makes use of stateof-the-art automatic speech recognition techniques. Our best system built for the CHiME-4 challenge consists on the combination of two different sub-systems in order to deal with the variety of acoustic conditions. Each sub-system in turn, follows a hybrid approach with different acoustic models, such as Deep Neural Networks or BLSTM Networks., The work leading to this invention has received funding from the European Union Seventh Framework Programme (FP7/2007- 2013) under grant agreement no 287755 (transLectures). Also, it has received funding from the EU¿s ICT Policy Support Programme as part of the Competitiveness and Innovation Framework Programme under grant agreement no 621030 (EMMA). In addition, this work has been supported by the Spanish research project MORE TIN2015-68326-R (MINECO/FEDER) and the Spanish Government with the FPU scholarship FPU13/06241.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Contribución de congreso (ConferenceOutput)

Fiber-distributed Signal Processing: Where the Space Dimension Comes into Play

  • Gasulla Mestre, Ivana|||0000-0001-8088-7796
  • García-Cortijo, Sergi|||0000-0002-7237-9457
  • Barrera Vilar, David|||0000-0002-1700-6842
  • Sales Maicas, Salvador|||0000-0001-9457-976X
  • Hervás-Peralta, Javier
[EN] We implement "fiber-distributed signal processing" in the context of fiber-wireless communications by using different multicore fibers, providing both radio access distribution and microwave photonics signal processing in the same fiber medium., This research was supported by the ERC Consolidator Grant 724663, the Spanish Projects TEC2014-60378-C2-1-R, TEC2015-62520-ERC and TEC2016-80150-R, the Spanish scholarships MECD FPU13/04675 for J. Hervás and MINECO BES-2015-073359 for S. García, and Spanish MINECO Ramón y Cajal RYC-2014-16247 for I. Gasulla
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Contribución de congreso (ConferenceOutput)

The MLLP-UPV German-English Machine Translation System for WMT18

  • Iranzo-Sánchez, Javier
  • Baquero-Arnal, Pau
  • Martínez-Villaronga, Adrià
  • Garcés Díaz-Munío, Gonçal|||0000-0002-2594-5858
  • Civera Saiz, Jorge|||0000-0002-0963-0143
  • Juan, Alfons|||0000-0002-9984-4072
[EN] This paper describes the statistical machine translation system built by the MLLP research group of Universitat Politècnica de València for the German¿English news translation shared task of the EMNLP 2018 Third Conference on Machine Translation (WMT18). We used an ensemble of Transformer architecture¿based neural machine translation systems. To train our system under ¿constrained¿ conditions, we filtered the provided parallel data with a scoring technique using character-based language models, and we added parallel data based on synthetic source sentences generated from the provided monolingual corpora., The research leading to these results has received funding from the European Union's Horizon 2020 research and innovation programme under grant agreement no. 761758 (X5gon); the Spanish government's TIN2015-68326-R (MINECO/FEDER) research project MORE, university collaboration grant programme 2017-2018, and faculty training scholarship FPU13/06241; the Generalitat Valenciana's predoctoral research scholarship ACIF/2017/055; as well as the Universitat Politecnica de València's PAID-01-17 R&D support programme.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Contribución de congreso (ConferenceOutput)

Animal Ludens: Building Intelligent Playful Environments for Animals

  • Pons Tomás, Patricia
  • Catalá Bolós, Alejandro
  • Jaén Martínez, Francisco Javier|||0000-0002-8815-9643
Looking for effective ways to understand how animals interact with computer-mediated systems, Animal-Computer Interaction (ACI) research should rely on the most natural and intrinsic behavior among the majority of living species: play. Animals are naturally motivated towards playing. Playful environments are, therefore, a promising scenario in which to start developing animal-centered ecosystems, and there are plenty of circumstances where playful environments could help to improve animals well-being. However, developing a custom system for each possible context remains unfeasible, and more appealing solutions are required. If playful environments were equipped with intelligent capabilities, they could learn from the animals behavior and automatically adapt themselves to the animals needs and preferences by creating engaging playful activities for different purposes. Hence, this work will define intelligent playful environments for animals and explain how Ambient Intelligence (AmI) can contribute to create adaptable playful experiences for animals in order to improve their quality of life., This work was partially funded by the Spanish Ministry of Science and Innovation under the National R&D&I Program within the project CreateWorlds (TIN2010-20488). It also received support from a postdoctoral fellowship within the VALi+d Program of the Conselleria d'Educació, Cultura i Esport (Generalitat Valenciana) awarded to Alejandro Catalá (APOSTD/2013/013). The work of Patricia Pons has been supported by the Universitat Politecnica de Valencia under the 'Beca de Excelencia" program, and currently by an FPU fellowship from the Spanish Ministry of Education, Culture and Sports (FPU13/03831).
RUC. Repositorio da Universidade da Coruña
Publicación Contribución de congreso (ConferenceOutput)

EN-ES-CS: An English-Spanish Code-Switching Twitter Corpus for Multilingual Sentiment Analysis

  • Vilares, David
  • Alonso, Miguel A.
  • Gómez-Rodríguez, Carlos
[Abstract]: Code-switching texts are those that contain terms in two or more different languages, and they appear increasingly often in social media. The aim of this paper is to provide a resource to the research community to evaluate the performance of sentiment classification techniques on this complex multilingual environment, proposing an English-Spanish corpus of tweets with code-switching (EN-ES-CS CORPUS). The tweets are labeled according to two well-known criteria used for this purpose: SentiStrength and a trinary scale (positive, neutral and negative categories). Preliminary work on the resource is already done, providing a set of baselines for the research community., This research is supported by the Ministerio de Economía y Competitividad (FFI2014-51978-C2) and Xunta de Galicia (R2014/034). David Vilares is funded by the Ministerio de Educación, Cultura y Deporte (FPU13/01180). Carlos Gómez-Rodríguez is funded by an Oportunius program grant., Xunta de Galicia; R2014/034
RUC. Repositorio da Universidade da Coruña
Publicación Contribución de congreso (ConferenceOutput)

LyS at TASS 2014: A Prototype for Extracting and Analysing Aspects from Spanish tweets, LyS en TASS 2014: Un prototipo para la extracción y análisis de aspectos en tuits.

  • Vilares, David
  • Doval, Yerai
  • Alonso, Miguel A.
  • Gómez-Rodríguez, Carlos
[Abstract]: This paper describes our participation at the third edition of the work- shop on Sentiment Analysis focused on Spanish tweets, tass 2014. This year’s eval- uation campaign includes four challenges: (1) global sentiment analysis, (2) topic classification, (3) aspect-extraction and (4) aspect-based sentiment analysis. Tasks 1 and 2 are addressed from a machine learning approach, using several linguistic resources and other information extracted from the training corpus to feed to a su- pervised classifier. With respect to task 3, we develop a naive approach, collecting a set of representations to identify the predefined aspects requested by the organisers. Finally, task 4 uses heuristics to identify the scope of each aspect, to then classify their sentiment via a supervised classifier. The experimental results are promising and will serve us as the starting point to develop more complex techniques., [Resumen]: Este artículo describe nuestra participación en la tercera edición del taller de análisis del sentimiento de tuits escritos en castellano, el tass 2014. En la evaluación competitiva de este año, se han propuesto cuatro retos: (1) análisis del sentimiento a nivel global, (2) clasificación de tópicos, (3) extracción de aspectos y (4) análisis del sentimiento a nivel aspectual. Para las tareas 1 y 2 empleamos una aproximación basada en aprendizaje automático, donde distintos recursos lingüísticos e información extraída del conjunto de entrenamiento son utilizados para entrenar un clasificador supervisado. Para abordar la tarea 3, nuestra aproximación recolecta una lista de representaciones que es empleada para identificar los aspectos requeridos por los organizadores. Por último, la tarea 4 delega en heurísticas para identificar el alcance de cada aspecto, para después determinar su sentimiento a través de un clasificador supervisado. Los resultados experimentales son prometedores y nos servirán para desarrollar técnicas más complejas en el futuro., Research reported in this paper has been partially funded by Ministerio de Educación, Cultura y Deporte (FPU13/01180) and Xunta de Galicia (Grant CN2012/008)., Xunta de Galicia; CN2012/008
RUC. Repositorio da Universidade da Coruña
Publicación Contribución de congreso (ConferenceOutput)

LyS at TASS 2015: Deep Learning Experiments for Sentiment Analysis on Spanish Tweets

  • Vilares, David
  • Doval, Yerai
  • Alonso, Miguel A.
  • Gómez-Rodríguez, Carlos
[Abstract]: This paper describes the participation of the LyS group at tass 2015. In this year’s edition, we used a long short-term memory neural network to address the two proposed challenges: (1) sentiment analysis at a global level and (2) aspect-based sentiment analysis on football and political tweets. The performance of this deep learning approach is compared to our last-year model, based on a square-regularized logistic regression. Experimental results show that strategies such as unsupervised pre-training, sentiment-specific word embedding or modifying the current architec- ture might be needed to achieve state-of-the-art results., [Resumen]: Este art´ıculo describe la participaci´on del grupo LyS en el tass 2015. En la edici´on de este a˜no, hemos utilizado una red neuronal denominada long short- term memory para abordar los dos retos propuestos: (1) an´alisis del sentimiento a nivel global y (2) an´alisis del sentimiento a nivel de aspectos sobre tuits futbol´ısticos y de pol´ıtica. El rendimiento obtenido por esta red de aprendizaje profundo es comparado con el de nuestro sistema del a˜no pasado, una regresi´on log´ıstica con una regularizaci´on cuadr´atica. Los resultados experimentales muestran que es necesario incluir estrategias como pre-entrenamiento no supervisado, t´ecnicas espec´ıficas para representar palabras como vectores o modificar la arquitectura actual para alcanzar resultados acordes con el estado del arte., This research is supported by the Ministerio de Economía y Competitividad y FEDER (FFI2014-51978-C2) and Xunta de Galicia (R2014/034). The first author is funded by the Ministerio de Educación, Cultura y Deporte (FPU13/01180)., Xunta de Galicia; R2014/034
RUC. Repositorio da Universidade da Coruña
Publicación Contribución de congreso (ConferenceOutput)

Identificación Automática del Idioma en Twitter: Adaptación de Identificadores del Estado del Arte al Contexto Ibérico, Automatic Language Identification in Twitter: Adapting State-of-the-Art Identifiers to the Iberian Context

  • Doval, Yerai
  • Vilares, David
  • Vilares, Jesús
[Abstract]: We describe here our partipation in TweetLID. After having studied the problem of language identification, the resources available, and designed a text conflation approach for this kind of tasks, we joined the competition with two sys- tems: the first one was based in the guesser langdetect, re-trained and adapted in order to work with conflated text; the second one was an approach based on majo- rity vote which used a set of re-trained and adapted classifiers. Results are analized both globally and at language and tweet-type levels., [Resumen]: Describimos aquí nuestra participación en el TweetLID. Tras estudiar la problemática de la identificación del idioma, los recursos disponibles y diseñar una solución para la normalización del texto en este tipo de tareas, presentamos dos sis- temas a competición: el primero basado en el clasificador langdetect, re-entrenado y adaptado para trabajar con texto normalizado; el segundo, una solución basada en la votación de clasificadores individuales igualmente re-entrenados y adaptados. Los resultados son analizados tanto globalmente como por idioma y tipo de tuit., Trabajo parcialmente financiado por el Ministeriode Economía y Competitividad y FEDER (TIN2010-18552-C03-02), Ministerio de Educación, Cultura y Deporte (Beca FPU ref. 13/01180) y Xunta de Galicia (CN2012/008)., Xunta de Galicia; CN2012/008
RUC. Repositorio da Universidade da Coruña
Publicación Contribución de congreso (ConferenceOutput)

Detecting Perspectives in Political Debates

  • Vilares, David
  • He, Yulan
[Abstract]: We explore how to detect people’s perspectives that occupy a certain proposition. We propose a Bayesian modelling approach where topics (or propositions) and their associated perspectives (or viewpoints) are modeled as latent variables. Words associated with topics or perspectives follow different generative routes. Based on the extracted perspectives, we can extract the top associated sentences from text to generate a succinct summary which allows a quick glimpse of the main viewpoints in a document. The model is evaluated on debates from the House of Commons of the UK Parliament, revealing perspectives from the debates without the use of labelled data and obtaining better results than previous related solutions under a variety of evaluations., We thank Charles Marshall for crawling the HCD data. DV was funded by MECD (FPU 13/01180), MINECO (FFI2014-51978-C2-2-R) and Inditex-UDC grants for research stays. YH is partly funded by the Natural Science Foundation of China (61528302)., China. Natural Science Foundation of China; 61528302
RUC. Repositorio da Universidade da Coruña
Publicación Contribución de congreso (ConferenceOutput)

Modelado dinámico y simulación de un robot trepador tipo serie con 4 grados de libertad

  • Peidró, Adrián
  • Hortal, Alba
  • Gil, Arturo
  • Marín, José María
  • Úbeda, David
  • Reinoso, Óscar
[Resumen] Este artículo presenta el modelado dinámico completo de un robot trepador con arquitectura serie y 4 grados de libertad. En primer lugar se obtiene el modelo cinemático del robot, modelándolo como un sistema articulado libre en el espacio. A continuación, se obtiene su modelo dinámico inverso empleando las ecuaciones de Lagrange. El modelo obtenido permite resolver la dinámica inversa del robot trepador estudiado independientemente de cuál de sus garras se fije al entorno o de cuál sea su postura en dicho entorno. El modelo obtenido ha sido implementado en una interfaz gráfica de usuario que permite simular la cinemática y la dinámica inversa del robot estudiado., Este trabajo ha sido financiado por el Ministerio de Educación a través de la ayuda FPU13/00413, por el Ministerio de Economía a través del proyecto DPI2013-41557-P, y por la Generalitat Valenciana a través del proyecto AICO/2015/021, Generalitat Valenciana; AICO/2015/021, https://doi.org/10.17979/spudc.9788497498081
RUC. Repositorio da Universidade da Coruña
Publicación Contribución de congreso (ConferenceOutput)

Seguimiento y análisis automático de contenidos en redes sociales

  • Alonso, Miguel A.
  • Gómez-Rodríguez, Carlos
  • Vilares, David
  • Doval, Yerai
  • Vilares, Jesús
[Abstract]: La Minería de Opiniones es la disciplina que aborda el tratamiento automático de las opiniones contenidas en un texto. Permite, por ejemplo, determinar si en un texto se está opinando o no, o si la polaridad o sentimiento que se expresa en el mismo es positiva, negativa o mixta. También permite la extracción automática de características, lo que posibilita conocer la percepción que los autores tienen sobre aspectos concretos de un tema determinado. Este trabajo, tras realizar una introducción a dicho ámbito, presenta una aproximación propia al mismo, la cual destaca por emplear información sintáctica, así como por estar especialmente adaptada a uno de los contextos de trabajo más complicados, Twitter. Dicha tecnología es fácilmente aplicable a tareas de inteligencia., Investigación financiada por Ministerio de Economía y Competitividad (FFI2014-51978-C2-2-R), Xunta de Galicia (R2014/034) y Ministerio de Educación, Cultura y Deporte (FPU13/01180)., Xunta de Galicia; R2014/034
RUC. Repositorio da Universidade da Coruña
Publicación Contribución de congreso (ConferenceOutput)

A non-projective greedy dependency parser with bidirectional LSTMs

  • Vilares, David
  • Gómez-Rodríguez, Carlos
[Abstract]: The LyS-FASTPARSE team present BIST-COVINGTON, a neural implementation of the Covington (2001) algorithm for non-projective dependency parsing. The bidirectional LSTM approach by Kiperwasser and Goldberg (2016) is used to train a greedy parser with a dynamic oracle to mitigate error propagation. The model participated in the CoNLL 2017 UD Shared Task. In spite of not using any ensemble methods and using the baseline segmentation and PoS tagging, the parser obtained good results on both macro-average LAS and UAS in the big treebanks category (55 languages), ranking 7th out of 33 teams. In the all treebanks category (LAS and UAS) we ranked 16th and 12th. The gap between the all and big categories is mainly due to the poor performance on four parallel PUD treebanks, suggesting that some ‘suffixed’ treebanks (e.g. Spanish-AnCora) perform poorly on cross-treebank settings, which does not occur with the corresponding ‘unsuffixed’ treebank (e.g. Spanish). By changing that, we obtain the 11th best LAS among all runs (official and unofficial). The code is made available at https://github.com/CoNLL-UD-2017/LyS-FASTPARSE., David Vilares is funded by an FPU Grant 13/01180. Carlos Gómez-Rodríguez has received funding from the European Research Council (ERC), under the European Union’s Horizon 2020 research and innovation programme (FASTPARSE, grant agreement No 714150). Both authors have received funding from the TELEPARES-UDC project from MINECO.
RUC. Repositorio da Universidade da Coruña
Publicación Contribución de congreso (ConferenceOutput)

Sentiment Analysis on Monolingual, Multilingual and Code-Switching Twitter Corpora

  • Vilares, David
  • Alonso, Miguel A.
  • Gómez-Rodríguez, Carlos
[Abstract]: We address the problem of performing po- larity classification on Twitter over differ- ent languages, focusing on English and Spanish, comparing three techniques: (1) a monolingual model which knows the language in which the opinion is written, (2) a monolingual model that acts based on the decision provided by a language iden- tification tool and (3) a multilingual model trained on a multilingual dataset that does not need any language recognition step. Results show that multilingual models are even able to outperform the monolingual models on some monolingual sets. We introduce the first code-switching corpus with sentiment labels, showing the robust- ness of a multilingual approach., This research is supported by the Ministerio de Economía y Competitividad (FFI2014-51978-C2) and Xunta de Galicia (R2014/034). The first authoris funded by the Ministerio de Educación, Cultura y Deporte (FPU13/01180)., Xunta de Galicia; R2014/034
RUC. Repositorio da Universidade da Coruña
Publicación Contribución de congreso (ConferenceOutput)

LyS at SemEval-2016 Task 4: Exploiting Neural Activation Values for Twitter Sentiment Classification and Quantification

  • Vilares, David
  • Doval, Yerai
  • Alonso, Miguel A.
  • Gómez-Rodríguez, Carlos
[Abstract]: In this paper we describe our deep learning approach for solving both two-, three- and fiveclass tweet polarity classification, and twoand five-class quantification. We first trained a convolutional neural network using pretrained Twitter word embeddings, so that we could extract the hidden activation values from the hidden layers once some input had been fed to the network. These values were then used as features for a support vector machine in both the classification and quantification subtasks, together with additional linguistic information in the former scenario. The results obtained for the classification subtasks show that this approach performs better than a single convolutional network, and for the quantification part it also yields good results. Official rankings locate us: 2nd (practically tied with 1st) for the binary classification task, 2nd for binary quantification and 4th (practically tied with 3rd) for the five-class polarity classification challenge., This research is supported by the Ministerio de Economía y Competitividad (FFI2014-51978-C2). David Vilares is funded by the Ministerio de Educación, Cultura y Deporte (FPU13/01180). Yerai Doval is funded by the Ministerio de Economía y Competitividad (BES-2015-073768). Carlos Gómez-Rodríguez is funded by an Oportunius program grant (Xunta de Galicia)
RUC. Repositorio da Universidade da Coruña
Publicación Contribución de congreso (ConferenceOutput)

Nonparametric Inference in Mixture Cure Models

  • López-Cheda, Ana
  • Cao, Ricardo
  • Jácome, M. A.
  • Keilegom, Ingrid Van
Presented at the XoveTIC Congress, A Coruña, Spain, 27–28 September 2018., [Abstract]: A completely nonparametric method for the estimation of mixture cure models is proposed. Nonparametric estimators for the cure probability (incidence) and for the survival function of the uncured population (latency) are introduced. In addition, a bootstrap bandwidth selection method for each nonparametric estimator is considered. The methodology is applied to a dataset of colorectal cancer patients from the University Hospital of A Coruña (CHUAC). Furthermore, a nonparametric covariate significance test for the incidence is proposed. The test is extended to non-continuous covariates: binary, discrete and qualitative, and also to contexts with a large number of covariates. The method is applied to a sarcomas dataset from the University Hospital of Santiago (CHUS)., The first author was sponsored by the Spanish FPU (Formación de Profesorado Universitario) Grant from MECD (Ministerio de Educación, Cultura y Deporte) with reference FPU13/01371. The work has been partially carried out during two visits at the Université catholique de Louvain. The first stay was financed by INDITEX and the second one was supported by the research group MODES (Modelización, Optimización e Inferencia Estadística). All the authors acknowledge partial support by the MICINN (Ministerio de Ciencia e Innovación) Grant MTM2011-22392 and the MINECO (Ministerio de Economía y Competitividad) Grant MTM2014-52876-R. The first three authors acknowledge partial support of Xunta de Galicia (Grupos de Referencia Competitiva CN2012/130, ED431C-2016-015 and Centro Singular de Investigación de Galicia ED431G/01), all of them through the ERDF (European Regional Development Fund). Financial support from the European Research Council (2016-2021, Horizon 2020 / ERC grant agreement No. 694409) is gratefully acknowledged., Xunta de Galicia; CN2012/130, Xunta de Galicia; ED431C-2016-015, Xunta de Galicia; ED431G/01
Zaguán. Repositorio Digital de la Universidad de Zaragoza
Publicación Contribución de congreso (ConferenceOutput)

Quantification of sprouting angiogenesis under the effect of different growth factors involved in the tumor microenvironmen

  • Del Amo, C.
  • Borau, C.
  • Asín, J.
  • García-Aznar, J. M.
One of the most important problems in tumor control is the management of metastatic process. Angiogenesis or the formation of new blood vessels from preexisting ones plays a crucial role in the expansion of the tumor by providing oxygen, nutrition and conduits for cancer cells to invade and metastasize new tissues¹. Abnormalities of growth factors (GFs) released such as PDGFs (Platelet Derived Growth Factor) could be involved in malignant human diseases2,3. Inflammation and cancer present similar mechanisms of development including angiogenesis or cell proliferation4. In order to know the effect on sprouting promotion of GFs existent in the tumor environment such as VEGF (Vascular Endothelial Growth Factor), PDGF, BMP2 (Bone Morphogenetic Protein 2) or TGF-ß (Transforming Growth Factor-ß), we have developed a microfluidic-based test based on devices designed by Farahat et al. (2012)5, which allows to the user the quantification of sprouting formation under the effect of these GFs. TGF-ß pathway involved in tumor progression in multiple human cancers, instigates phenotypical changes affecting to the cell growth, differentiation and migration6. Knowing the overexpression of GFs such as VEGF or BMP2 in tumors7,8, we aimed to compare its effect on endothelial cells in angiogenesis. Analyzing the promotion of sprout in normal conditions under GFs addition would be possible to determine which of these molecules could decrease or promote the advance of the endothelial cells. The results obtained in this work indicated that VEGF is the most important factor to enhance the angiogenic process while non-specific factors such as BMP2 or TGF-ß show a low effectiveness. In the case of PDGF, the negative effect of this molecule observed in our assays could be explained by the non-optimal balance of concentration. Furthermore, we are currently working to quantify the effect of fluid flow on the sprouting promotion.
Recercat. Dipósit de la Recerca de Catalunya
Publicación Contribución de congreso (ConferenceOutput)

From emotions to action units with hidden and semi-hidden-task learning

  • Ruiz, Adrià
  • Van de Weijer, Joost
  • Binefa i Valls, Xavier
Comunicació presentada a: IEEE International Conference on Computer Vision (ICCV 2015) celebrat a Santiago, Xile, de l'11 al 18 de desembre de 2015., Limited annotated training data is a challenging problem in Action Unit recognition. In this paper, we investigate how the use of large databases labelled according to the 6 universal facial expressions can increase the generalization ability of Action Unit classifiers. For this purpose, we propose a novel learning framework: Hidden-Task Learning. HTL aims to learn a set of Hidden-Tasks (Action Units) for which samples are not available but, in contrast, training data is easier to obtain from a set of related Visible- Tasks (Facial Expressions). To that end, HTL is able to exploit prior knowledge about the relation between Hidden and Visible-Tasks. In our case, we base this prior knowledge on empirical psychological studies providing statistical correlations between Action Units and universal facial expressions. Additionally, we extend HTL to Semi-Hidden Task Learning (SHTL) assuming that Action Unit training samples are also provided. Performing exhaustive experiments over four different datasets, we show that HTL and SHTL improve the generalization ability of AU classifiers by training them with additional facial expression data. Additionally, we show that SHTL achieves competitive performance compared with state-of-the-art Transductive Learning approaches which face the problem of limited training data by using unlabelled test samples during training., This paper is part of a project that has received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement No 645012. Adria Ruiz and Xavier Binefa would also like to acknowledge Spanish Government to provide support under grants CICYT TIN2012-39203 and FPU13/01740. Joost van de Weijer acknowledges Project TIN2013-41751 of the Spanish Ministry of Science and the Generalitat de Catalunya Project under Grant 2014-SGR-221.
Academica-e. Repositorio Institucional de la Universidad Pública de Navarra
Publicación Contribución de congreso (ConferenceOutput)

Hydrogen-based energy storage for a distributed generation system

  • San Martín Biurrun, Idoia
  • Berrueta Irigoyen, Alberto
  • Ursúa Rubio, Alfredo
  • Sanchis Gúrpide, Pablo
Incluye póster., One of the most typical distributed generation systems are electrical microgrid, which consist on small electrical grids, generally connected to the main grid, with a decentralized management structure. Electrical microgrids allow higher renewable energy integration in the grid, achieving a cost decrease and improving the grid quality [1]. These mi- crogrids incorporate renewable generation systems and energy consumers. Moreover, they have storage systems to balance generation and consumption as well as the exchanged power with the main grid. Traditionally, lead-acid batter- ies have been used in microgrids. However, these batteries have some drawbacks, being the most important its poor performance in partial state of charge, which is critical for a microgrid. A suitable option for the storage system is hy- drogen technology. These systems have high energy density, which makes the storage system able to assume seasonal variability of renewable resources. This paper proposes a sizing methodology for storage systems based on hydrogen for grid-tied electrical microgrids. This methodology optimizes the relationship between the storage system size and the consumption of grid power., Spanish Ministry of Economy and Competitiveness under grant DPI2013-42853-R; 2015 Annual Grants for Research and Development of Fundación Bancaria Caja Navarra (supporting project 70276); FPU Program of the Spanish Ministry of Education, Culture and Sport (FPU13/00542).