Efficient Monte Carlo methods for multi-dimensional learning with classifier chains

Multi-dimensional classification (MDC) is the supervised learning problem where an instance is associated with multiple classes, rather than with a single class, as in traditional classification problems. Since these classes are often strongly correlated, modeling the dependencies between them allows MDC methods to improve their performance – at the expense of an increased computational cost. In this paper we focus on the classifier chains (CC) approach for modeling dependencies, one of the most popular and highest-performing methods for multi-label classification (MLC), a particular case of MDC which involves only binary classes (i.e., labels). The original CC algorithm makes a greedy approximation, and is fast but tends to propagate errors along the chain. Here we present novel Monte Carlo schemes, both for finding a good chain sequence and performing efficient inference. Our algorithms remain tractable for high-dimensional data sets and obtain the best predictive performance across several real data sets.

Real-Time Adaptive Multi-Classifier Multi-Resolution Visual Tracking Framework for Unmanned Aerial Vehicles

This paper presents a novel robust visual tracking framework, based on discriminative method, for Unmanned Aerial Vehicles (UAVs) to track an arbitrary 2D/3D target at real-time frame rates, that is called the Adaptive Multi-Classifier Multi-Resolution (AMCMR) framework. In this framework, adaptive Multiple Classifiers (MC) are updated in the (k-1)th frame-based Multiple Resolutions (MR) structure with compressed positive and negative samples, and then applied them in the kth frame-based Multiple Resolutions (MR) structure to detect the current target. The sample importance has been integrated into this framework to improve the tracking stability and accuracy. The performance of this framework was evaluated with the Ground Truth (GT) in different types of public image databases and real flight-based aerial image datasets firstly, then the framework has been applied in the UAV to inspect the Offshore Floating Platform (OFP). The evaluation and application results show that this framework is more robust, efficient and accurate against the existing state-of-art trackers, overcoming the problems generated by the challenging situations such as obvious appearance change, variant illumination, partial/full target occlusion, blur motion, rapid pose variation and onboard mechanical vibration, among others. To our best knowledge, this is the first work to present this framework for solving the online learning and tracking freewill 2D/3D target problems, and applied it in the UAVs.

Estimación eficiente de atributos demográficos del rostro humano en imágenes

Sin duda, el rostro humano ofrece mucha más información de la que pensamos. La cara transmite sin nuestro consentimiento señales no verbales, a partir de las interacciones faciales, que dejan al descubierto nuestro estado afectivo, actividad cognitiva, personalidad y enfermedades. Estudios recientes [OFT14, TODMS15] demuestran que muchas de nuestras decisiones sociales e interpersonales derivan de un previo análisis facial de la cara que nos permite establecer si esa persona es confiable, trabajadora, inteligente, etc. Esta interpretación, propensa a errores, deriva de la capacidad innata de los seres humanas de encontrar estas señales e interpretarlas. Esta capacidad es motivo de estudio, con un especial interés en desarrollar métodos que tengan la habilidad de calcular de manera automática estas señales o atributos asociados a la cara. Así, el interés por la estimación de atributos faciales ha crecido rápidamente en los últimos años por las diversas aplicaciones en que estos métodos pueden ser utilizados: marketing dirigido, sistemas de seguridad, interacción hombre-máquina, etc. Sin embargo, éstos están lejos de ser perfectos y robustos en cualquier dominio de problemas. La principal dificultad encontrada es causada por la alta variabilidad intra-clase debida a los cambios en la condición de la imagen: cambios de iluminación, oclusiones, expresiones faciales, edad, género, etnia, etc.; encontradas frecuentemente en imágenes adquiridas en entornos no controlados. Este de trabajo de investigación estudia técnicas de análisis de imágenes para estimar atributos faciales como el género, la edad y la postura, empleando métodos lineales y explotando las dependencias estadísticas entre estos atributos. Adicionalmente, nuestra propuesta se centrará en la construcción de estimadores que tengan una fuerte relación entre rendimiento y coste computacional. Con respecto a éste último punto, estudiamos un conjunto de estrategias para la clasificación de género y las comparamos con una propuesta basada en un clasificador Bayesiano y una adecuada extracción de características. Analizamos en profundidad el motivo de porqué las técnicas lineales no han logrado resultados competitivos hasta la fecha y mostramos cómo obtener rendimientos similares a las mejores técnicas no-lineales. Se propone un segundo algoritmo para la estimación de edad, basado en un regresor K-NN y una adecuada selección de características tal como se propuso para la clasificación de género. A partir de los experimentos desarrollados, observamos que el rendimiento de los clasificadores se reduce significativamente si los ´estos han sido entrenados y probados sobre diferentes bases de datos. Hemos encontrado que una de las causas es la existencia de dependencias entre atributos faciales que no han sido consideradas en la construcción de los clasificadores. Nuestro resultados demuestran que la variabilidad intra-clase puede ser reducida cuando se consideran las dependencias estadísticas entre los atributos faciales de el género, la edad y la pose; mejorando el rendimiento de nuestros clasificadores de atributos faciales con un coste computacional pequeño. Abstract Surely the human face provides much more information than we think. The face provides without our consent nonverbal cues from facial interactions that reveal our emotional state, cognitive activity, personality and disease. Recent studies [OFT14, TODMS15] show that many of our social and interpersonal decisions derive from a previous facial analysis that allows us to establish whether that person is trustworthy, hardworking, intelligent, etc. This error-prone interpretation derives from the innate ability of human beings to find and interpret these signals. This capability is being studied, with a special interest in developing methods that have the ability to automatically calculate these signs or attributes associated with the face. Thus, the interest in the estimation of facial attributes has grown rapidly in recent years by the various applications in which these methods can be used: targeted marketing, security systems, human-computer interaction, etc. However, these are far from being perfect and robust in any domain of problems. The main difficulty encountered is caused by the high intra-class variability due to changes in the condition of the image: lighting changes, occlusions, facial expressions, age, gender, ethnicity, etc.; often found in images acquired in uncontrolled environments. This research work studies image analysis techniques to estimate facial attributes such as gender, age and pose, using linear methods, and exploiting the statistical dependencies between these attributes. In addition, our proposal will focus on the construction of classifiers that have a good balance between performance and computational cost. We studied a set of strategies for gender classification and we compare them with a proposal based on a Bayesian classifier and a suitable feature extraction based on Linear Discriminant Analysis. We study in depth why linear techniques have failed to provide competitive results to date and show how to obtain similar performances to the best non-linear techniques. A second algorithm is proposed for estimating age, which is based on a K-NN regressor and proper selection of features such as those proposed for the classification of gender. From our experiments we note that performance estimates are significantly reduced if they have been trained and tested on different databases. We have found that one of the causes is the existence of dependencies between facial features that have not been considered in the construction of classifiers. Our results demonstrate that intra-class variability can be reduced when considering the statistical dependencies between facial attributes gender, age and pose, thus improving the performance of our classifiers with a reduced computational cost.

Fenología y capacidad de dispersión de Monochamus galloprovincialis (Olivier 1795) en la Península Ibérica = Phenology and dispersal ability of Monochamus galloprovincialis (Olivier 1795) at Iberian Peninsula

La aparición y avance de la enfermedad del marchitamiento del pino (Pine Wilt Desease, PWD), causada por Bursaphelenchus xylophilus (Nematoda; Aphelenchoididae), el nematodo de la madera del pino (NMP), en el suroeste de Europa, ha puesto de manifiesto la necesidad de estudiar la fenología y la dispersión de su único vector conocido en Europa, Monochamus galloprovincialis (Col., Cerambycidae). El análisis de 12 series de emergencias entre 2010 y 2014, registradas en Palencia, València y Teruel, con material procedente de diversos puntos de la península ibérica, demostró una alta variabilidad en la fenología de M. galloprovincialis y la divergencia térmica respecto de las poblaciones portuguesas. Para éstas, el establecimiento de los umbrales térmicos de desarrollo de las larvas post-dormantes del vector (12,2 y 33,5ºC) permitió la predicción de la emergencia mediana para la fecha en la que se acumulaban de 822 grados-día. Ninguna de las series analizadas en este trabajo necesitó de dichos grados-día estimados para la emergencia mediana. Asimismo, la emergencia se adelantó en las regiones más calurosas, mientras que se retrasó en las zonas más templadas. Más allá de la posible variabilidad entre poblaciones locales peninsulares, se detectaron indicios de que la diferencia en la acumulación de calor durante el otoño puede afectar el grado de maduración de las larvas invernantes, y su posterior patrón temporal de emergencia. Por último, también fueron observados comportamientos de protandria en las emergencias. Respecto a la fenología de su vuelo, entre los años 2010 y 2015, fueron ejecutados un total de 8 experimentos de captura de M. galloprovincialis mediante trampas cebadas con atrayentes en diferentes regiones (Castellón, Teruel, Segovia y Alicante) permitiendo el seguimiento del periodo de vuelo. Su análisis permitió constatar la disminución de las capturas y el acortamiento del periodo de vuelo con la altitud, el inicio del vuelo en el mes de mayo/junio a partir de los 14ºC de temperatura media diaria, la influencia de las altas temperaturas en la disminución de las capturas estivales (potencial causante de perfiles bimodales en las curvas de vuelo en las zonas menos frías), la evolución de la proporción de sexos a lo largo del periodo de vuelo (que muestra una mayor captura de hembras al inicio y de machos al final) y el comportamiento diurno y ligado a las altas temperaturas del vuelo circadiano del insecto. Dos redes de muestreo sistemático de insectos saproxílicos instaladas en la Comunitat Valencia (Red MUFFET, 15 parcelas, año 2013) y en Murcia (Red ESFP, 20 parcelas, años 2008-2010) permitieron el estudio de la comunidad de insectos relacionada con M. galloprovincialis. Cada una de las parcelas contaba con una trampa cebada con atrayentes y una estación meteorológica. El registro de más de 250 especies de coleópteros saproxílicos demostró el potencial que tiene el empleo de redes de trampas vigía para la detección temprana de organismos exóticos, además de permitir la caracterización y evaluación de las comunidades de entomofauna útil, representando una de las mejores herramientas de la gestión integrada de plagas. En este caso, la comunidad de saproxílicos estudiada mostró ser muy homogénea respecto a la variación ambiental de las zonas de muestreo, y que pese a las pequeñas variaciones entre las comunidades de los diferentes ecosistemas, el rol que M. galloprovincialis desempeña en ellas a lo largo de todo el gradiente estudiado es el mismo. Con todo, el análisis mediante redes de interacción mostró su relevancia ecológica al actuar de conector entre los diferentes niveles tróficos. Por último, un total de 12 experimentos de marcaje-liberación-recaptura desarrollados entre 2009 y 2012 en Castellón, Teruel, Valencia y Murcia permitieron evaluar el comportamiento dispersivo de M. galloprovincialis. Las detecciones mediante trampas cebadas de los insectos liberados se dieron por lo menos 8 días después de la emergencia. La abundancia de población pareció relacionada con la continuidad, la naturalización de la masa, y con la afección previa de incendios. La dispersión no estuvo influida por la dirección ni la intensidad de los vientos dominantes. La abundancia de material hospedante (en lo referente a las variables de masa y a los índices de competencia) influyó en la captura del insecto en paisajes fragmentados, aunque la ubicación de las trampas optimizó el número de capturas cuando se ubicaron en el límite de la masa y en zonas visibles. Por último también se constató que M. galloprovincialis posee suficiente capacidad de dispersión como para recorrer hasta 1500 m/día, llegando a alcanzar distancias máximas de 13600m o de 22100 m. ABSTRACT The detection and expansion of the Pine Wilt Desease (PWD), caused by Bursaphelenchus xylophilus (Nematoda; Aphelenchoididae), Pine Wood Nematode (PWN), in southwestern Europe since 1999, has triggered off the study of the phenology and the dispersion of its unique vector in the continent, Monochamus galloprovincialis (Coleoptera, Cerambycidae). The analysis of 12 emergence series between 2010 and 2014 registered in Palencia, Teruel and Valencia (Spain), registered from field colonized material collected at several locations of the Iberian Peninsula, showed a high variability in the emergence phenology of M. galloprovincialis. In addition, these patterns showed a very acute thermal divergence regarding a development model fitted earlier in Portugal. Such model forecasted the emergence of 50% of M. galloprovincialis individuals in the Setúbal Peninsula (Portugal) when an average of 822 degree-days (DD) were reached, based on the accumulation of heat from the 1st of March until emergence and lower and upper thresholds of 12.2 ºC and 33,5 °C respectively. In our results, all analyzed series needed less than 822 DD to complete the 50% of the emergence. Also, emergency occurred earlier in the hottest regions, while it was delayed in more temperate areas. Beyond the possible variability between local populations, the difference in the heat accumulation during the fall season may have affected the degree of maturation of overwintering larvae, and subsequently, the temporal pattern of M. galloprovincialis emergences. Therefore these results suggest the need to differentiate local management strategies for the PWN vector, depending on the location, and the climatic variables of each region. Finally, protandrous emergence patterns were observed for M. galloprovincialis in most of the studied data-sets. Regarding the flight phenology of M. galloprovincialis, a total of 8 trapping experiments were carried out in different regions of the Iberian Peninsula (Castellón, Teruel, Segovia and Alicante) between 2010 and 2015. The use of commercial lures and traps allowed monitoring of the flight period of M. galloprovincialis. The analyses of such curves, helped confirming different aspects. First, a decline in the number of catches and a shortening of the flight period was observed as the altitude increased. Flight period was recorded to start in May / June when the daily average temperature went over 14 ° C. A significant influence of high temperatures on the decrease of catches in the summer was found in many occasions, which frequently lead to a bimodal profile of the flight curves in warm areas. The evolution of sex ratio along the flight period shows a greater capture of females at the beginning of the period, and of males at the end. In addition, the circadian response of M. galloprovincialis to lured traps was described for the first time, concluding that the insect is diurnal and that such response is linked to high temperatures. Two networks of systematic sampling of saproxylic insects were installed in the Region of Valencia (Red MUFFET, 15 plots, 2013) and Murcia (Red ICPF, 20 plots, 2008-2010). These networks, intended to serve the double purpose of early-detection and long term monitoring of the saproxylic beetle assemblies, allowed the study of insect communities related to M. galloprovincialis. Each of the plots had a trap baited with attractants and a weather station. The registration of almost 300 species of saproxylic beetles demonstrated the potential use of such trapping networks for the early detection of exotic organisms, while at the same time allows the characterization and evaluation of useful entomological fauna communities, representing one of the best tools for the integrated pest management. In this particular case, the studied community of saproxylic beetles was very homogeneous with respect to environmental variation of the sampling areas, and despite small variations between communities of different ecosystems, the role that M. galloprovincialis apparently plays in them across the studied gradient seems to be the same. However, the analysis through food-webs showed the ecological significance of M. galloprovincialis as a connector between different trophic levels. Finally, 12 mark-release-recapture experiments were carried out between 2009 and 2012 in Castellón, Teruel, Valencia and Murcia (Spain) with the aim to describe the dispersive behavior of M. galloprovincialis as well as the stand and landscape characteristics that could influence its abundance and dispersal. No insects younger than 8 days were caught in lured traps. Population abundance estimates from mark-release-recapture data, seemed related to forest continuity, naturalization, and to prior presence of forest fires. On the other hand, M. galloprovincialis dispersal was not found to be significantly influenced by the direction and intensity of prevailing winds. The abundance of host material, very related to stand characteristics and spacing indexes, influenced the insect abundance in fragmented landscapes. In addition, the location of the traps optimized the number of catches when they were placed in the edge of the forest stands and in visible positions. Finally it was also found that M. galloprovincialis is able to fly up to 1500 m / day, reaching maximum distances of up to 13600 m or 22100 m.

Co-aggregation ability of cell Wall components of Saccharomyces cerevisiae to pathogenic bacteria

Autoaggregation in bacteria is the phenomenon of aggregation between cells of the same strain, whereas coaggregation is due to aggregation occurring among different species. Aggregation ability of prebiotic bacteria is related to adhesion ability, which is a prerequisite for the colonization and protection of the gastrointestinal tract in all animal species; however, coaggregation ability of prebiotic bacteria offers a possibility of close interaction with pathogenic bacteria.