Object Recognition with Pictorial Structures

This thesis presents a statistical framework for object recognition. The framework is motivated by the pictorial structure models introduced by Fischler and Elschlager nearly 30 years ago. The basic idea is to model an object by a collection of parts arranged in a deformable configuration. The appearance of each part is modeled separately, and the deformable configuration is represented by spring-like connections between pairs of parts. These models allow for qualitative descriptions of visual appearance, and are suitable for generic recognition problems. The problem of detecting an object in an image and the problem of learning an object model using training examples are naturally formulated under a statistical approach. We present efficient algorithms to solve these problems in our framework. We demonstrate our techniques by training models to represent faces and human bodies. The models are then used to locate the corresponding objects in novel images.

A Statistical Image-Based Shape Model for Visual Hull Reconstruction and 3D Structure Inference

We present a statistical image-based shape + structure model for Bayesian visual hull reconstruction and 3D structure inference. The 3D shape of a class of objects is represented by sets of contours from silhouette views simultaneously observed from multiple calibrated cameras. Bayesian reconstructions of new shapes are then estimated using a prior density constructed with a mixture model and probabilistic principal components analysis. We show how the use of a class-specific prior in a visual hull reconstruction can reduce the effect of segmentation errors from the silhouette extraction process. The proposed method is applied to a data set of pedestrian images, and improvements in the approximate 3D models under various noise conditions are shown. We further augment the shape model to incorporate structural features of interest; unknown structural parameters for a novel set of contours are then inferred via the Bayesian reconstruction process. Model matching and parameter inference are done entirely in the image domain and require no explicit 3D construction. Our shape model enables accurate estimation of structure despite segmentation errors or missing views in the input silhouettes, and works even with only a single input view. Using a data set of thousands of pedestrian images generated from a synthetic model, we can accurately infer the 3D locations of 19 joints on the body based on observed silhouette contours from real images.

Probabilistic Independence Networks for Hidden Markov Probability Models

Graphical techniques for modeling the dependencies of randomvariables have been explored in a variety of different areas includingstatistics, statistical physics, artificial intelligence, speech recognition, image processing, and genetics.Formalisms for manipulating these models have been developedrelatively independently in these research communities. In this paper weexplore hidden Markov models (HMMs) and related structures within the general framework of probabilistic independencenetworks (PINs). The paper contains a self-contained review of the basic principles of PINs.It is shown that the well-known forward-backward (F-B) and Viterbialgorithms for HMMs are special cases of more general inference algorithms forarbitrary PINs. Furthermore, the existence of inference and estimationalgorithms for more general graphical models provides a set of analysistools for HMM practitioners who wish to explore a richer class of HMMstructures.Examples of relatively complex models to handle sensorfusion and coarticulationin speech recognitionare introduced and treated within the graphical model framework toillustrate the advantages of the general approach.

Time-series regression models to study the short-term effects of environmental factors on health

Time series regression models are especially suitable in epidemiology for evaluating short-term effects of time-varying exposures on health. The problem is that potential for confounding in time series regression is very high. Thus, it is important that trend and seasonality are properly accounted for. Our paper reviews the statistical models commonly used in time-series regression methods, specially allowing for serial correlation, make them potentially useful for selected epidemiological purposes. In particular, we discuss the use of time-series regression for counts using a wide range Generalised Linear Models as well as Generalised Additive Models. In addition, recently critical points in using statistical software for GAM were stressed, and reanalyses of time series data on air pollution and health were performed in order to update already published. Applications are offered through an example on the relationship between asthma emergency admissions and photochemical air pollutants

Statistical modeling on (α, +∞)

Observations in daily practice are sometimes registered as positive values larger then a given threshold α. The sample space is in this case the interval (α,+∞), α > 0, which can be structured as a real Euclidean space in different ways. This fact opens the door to alternative statistical models depending not only on the assumed distribution function, but also on the metric which is considered as appropriate, i.e. the way differences are measured, and thus variability

Statistical modeling on coordinates

This paper is a first draft of the principle of statistical modelling on coordinates. Several causes —which would be long to detail—have led to this situation close to the deadline for submitting papers to CODAWORK’03. The main of them is the fast development of the approach along the last months, which let appear previous drafts as obsolete. The present paper contains the essential parts of the state of the art of this approach from my point of view. I would like to acknowledge many clarifying discussions with the group of people working in this field in Girona, Barcelona, Carrick Castle, Firenze, Berlin, G¨ottingen, and Freiberg. They have given a lot of suggestions and ideas. Nevertheless, there might be still errors or unclear aspects which are exclusively my fault. I hope this contribution serves as a basis for further discussions and new developments

Probabilistic models to assist maintenance of multiple instruments

The paper discusses maintenance challenges of organisations with a huge number of devices and proposes the use of probabilistic models to assist monitoring and maintenance planning. The proposal assumes connectivity of instruments to report relevant features for monitoring. Also, the existence of enough historical registers with diagnosed breakdowns is required to make probabilistic models reliable and useful for predictive maintenance strategies based on them. Regular Markov models based on estimated failure and repair rates are proposed to calculate the availability of the instruments and Dynamic Bayesian Networks are proposed to model cause-effect relationships to trigger predictive maintenance services based on the influence between observed features and previously documented diagnostics

Option pricing in market models driven by telegraph processes with jumps

Esta tesis está dividida en dos partes: en la primera parte se presentan y estudian los procesos telegráficos, los procesos de Poisson con compensador telegráfico y los procesos telegráficos con saltos. El estudio presentado en esta primera parte incluye el cálculo de las distribuciones de cada proceso, las medias y varianzas, así como las funciones generadoras de momentos entre otras propiedades. Utilizando estas propiedades en la segunda parte se estudian los modelos de valoración de opciones basados en procesos telegráficos con saltos. En esta parte se da una descripción de cómo calcular las medidas neutrales al riesgo, se encuentra la condición de no arbitraje en este tipo de modelos y por último se calcula el precio de las opciones Europeas de compra y venta.

A proposal for the diagnosis of uncertain dynamic systems based on interval models

The performance of a model-based diagnosis system could be affected by several uncertainty sources, such as,model errors,uncertainty in measurements, and disturbances. This uncertainty can be handled by mean of interval models.The aim of this thesis is to propose a methodology for fault detection, isolation and identification based on interval models. The methodology includes some algorithms to obtain in an automatic way the symbolic expression of the residual generators enhancing the structural isolability of the faults, in order to design the fault detection tests. These algorithms are based on the structural model of the system. The stages of fault detection, isolation, and identification are stated as constraint satisfaction problems in continuous domains and solved by means of interval based consistency techniques. The qualitative fault isolation is enhanced by a reasoning in which the signs of the symptoms are derived from analytical redundancy relations or bond graph models of the system. An initial and empirical analysis regarding the differences between interval-based and statistical-based techniques is presented in this thesis. The performance and efficiency of the contributions are illustrated through several application examples, covering different levels of complexity.

Avaluació de la toxicitat de metalls pesants i arsènic en diferents models biològics

En aquest estudi, la toxicitat de diversos metalls pesants i l'arsènic va ser analitzada utilitzant diferents models biològics. En la primera part d'aquest treball, el bioassaig de toxicitat Microtox, el qual està basat en la variació de l'emissió lumínica del bacteri luminiscent Vibrio fischeri, va ser utilitzat per establir les corbes dosi-resposta de diferents elements tòxics com el Zn(II), Pb(II), Cu(II), Hg(II), Ag(I), Co(II), Cd(II), Cr(VI), As(V) i As(III) en solucions aquoses. Els experiments es varen portar a terme a pH 6.0 i 7.0 per tal de mostrar que el pH pot influir en la toxicitat final mesurada d'alguns metalls degut als canvis relacionats amb la seva especiació química. Es varen trobar diferents tipus de corbes dosi-resposta depenent del metall analitzat i el pH del medi. En el cas de l'arsènic, l'efecte del pH en la toxicitat de l'arsenat i l'arsenit es va investigar utilitzant l'assaig Microtox en un rang de pHs comprès entre pH 5.0 i 9.0. Els valors d'EC50 determinats per l'As(V) disminueixen, reflectint un augment de la toxicitat, a mesura que el pH de la solució augmenta mentre que, en el cas de l'As(III), els valors d'EC50 quasi bé no varien entre pH 6.0 i 8.0 i només disminueixen a pH 9.0. HAsO42- i H2AsO3- es varen definir com les espècies més tòxiques. Així mateix, una anàlisi estadística va revelar un efecte antagònic entre les espècies químiques d'arsenat que es troben conjuntament a pH 6.0 i 7.0. D'altra banda, els resultats de dos mètodes estadístics per predir la toxicitat i les possibles interaccions entre el Co(II), Cd(II), Cu(II), Zn(II) i Pb(II) en mescles binàries equitòxiques es varen comparar amb la toxicitat observada sobre el bacteri Vibrio fischeri. L'efecte combinat d'aquests metalls va resultar ser antagònic per les mescles de Co(II)-Cd(II), Cd(II)-Zn(II), Cd(II)-Pb(II) i Cu(II)-Pb(II), sinèrgic per Co(II)-Cu(II) i Zn(II)-Pb(II) i additiu en els altres casos, revelant un patró complex de possibles interaccions. L'efecte sinèrgic de la combinació Co(II)-Cu(II) i la forta disminució de la toxicitat del Pb(II) quan es troba en presència de Cd(II) hauria de merèixer més atenció quan s'estableixen les normatives de seguretat ambiental. La sensibilitat de l'assaig Microtox també va ser determinada. Els valors d'EC20, els quals representen la toxicitat llindar mesurable, varen ser determinats per cada element individualment i es va veure que augmenten de la següent manera: Pb(II) < Ag(I) < Hg(II)  Cu(II) < Zn(II) < As(V) < Cd(II)  Co(II) < As(III) < Cr(VI). Aquests valors es varen comparar amb les concentracions permeses en aigues residuals industrials establertes per la normativa oficial de Catalunya (Espanya). L'assaig Microtox va resultar ser suficientment sensible per detectar els elements assajats respecte a les normes oficials referents al control de la contaminació, excepte en el cas del cadmi, mercuri, arsenat, arsenit i cromat. En la segona part d'aquest treball, com a resultats complementaris dels resultats previs obtinguts utilitzant l'assaig de toxicitat aguda Microtox, els efectes crònics del Cd(II), Cr(VI) i As(V) es varen analitzar sobre la taxa de creixement i la viabilitat en el mateix model biològic. Sorprenentment, aquests productes químics nocius varen resultar ser poc tòxics per aquest bacteri quan es mesura el seu efecte després de temps d'exposició llargs. Tot i això, en el cas del Cr(VI), l'assaig d'inhibició de la viabilitat va resultar ser més sensible que l'assaig de toxicitat aguda Microtox. Així mateix, també va ser possible observar un clar fenomen d'hormesis, especialment en el cas del Cd(II), quan s'utilitza l'assaig d'inhibició de la viabilitat. A més a més, diversos experiments es varen portar a terme per intentar explicar la manca de toxicitat de Cr(VI) mostrada pel bacteri Vibrio fischeri. La resistència mostrada per aquest bacteri podria ser atribuïda a la capacitat d'aquest bacteri de convertir el Cr(VI) a la forma menys tòxica de Cr(III). Es va trobar que aquesta capacitat de reducció depèn de la composició del medi de cultiu, de la concentració inicial de Cr(VI), del temps d'incubació i de la presència d'una font de carboni. En la tercera part d'aquest treball, la línia cel·lular humana HT29 i cultius primaris de cèl·lules sanguínies de Sparus sarba es varen utilitzar in vitro per detectar la toxicitat llindar de metalls mesurant la sobreexpressió de proteines d'estrès. Extractes de fangs precedents de diverses plantes de tractament d'aigues residuals i diferents metalls, individualment o en combinació, es varen analitzar sobre cultius cel·lulars humans per avaluar el seu efecte sobre la taxa de creixement i la capacitat d'induir la síntesi de les proteïnes Hsp72 relacionades amb l'estrès cel·lular. No es varen trobar efectes adversos significatius quan els components s'analitzen individualment. Nogensmenys, quan es troben conjuntament, es produeix un afecte advers sobre tan la taxa de creixement com en l'expressió de proteins d'estrès. D'altra banda, cèl·lules sanguínies procedents de Sparus sarba es varen exposar in vitro a diferents concentracions de cadmi, plom i crom. La proteïna d'estrès HSP70 es va sobreexpressar significativament després de l'exposició a concentracions tan febles com 0.1 M. Sota les nostres condicions de treball, no es va evidenciar una sobreexpressió de metal·lotioneïnes. Nogensmenys, les cèl·lules sanguínies de peix varen resultar ser un model biològic interessant per a ser utilitzat en anàlisis de toxicitat. Ambdós models biològics varen resultar ser molt adequats per a detectar acuradament la toxicitat produïda per metalls. En general, l'avaluació de la toxicitat basada en l'anàlisi de la sobreexpressió de proteïnes d'estrès és més sensible que l'avaluació de la toxicitat realitzada a nivell d'organisme. A partir dels resultats obtinguts, podem concloure que una bateria de bioassaigs és realment necessària per avaluar acuradament la toxicitat de metalls ja que existeixen grans variacions entre els valors de toxicitat obtinguts emprant diferents organismes i molts factors ambientals poden influir i modificar els resultats obtinguts.

Distributional semantics and machine learning for statistical machine translation

[EU]Lan honetan semantika distribuzionalaren eta ikasketa automatikoaren erabilera aztertzen dugu itzulpen automatiko estatistikoa hobetzeko. Bide horretan, erregresio logistikoan oinarritutako ikasketa automatikoko eredu bat proposatzen dugu hitz-segiden itzulpen- probabilitatea modu dinamikoan modelatzeko. Proposatutako eredua itzulpen automatiko estatistikoko ohiko itzulpen-probabilitateen orokortze bat dela frogatzen dugu, eta testuinguruko nahiz semantika distribuzionaleko informazioa barneratzeko baliatu ezaugarri lexiko, hitz-cluster eta hitzen errepresentazio bektorialen bidez. Horretaz gain, semantika distribuzionaleko ezagutza itzulpen automatiko estatistikoan txertatzeko beste hurbilpen bat lantzen dugu: hitzen errepresentazio bektorial elebidunak erabiltzea hitz-segiden itzulpenen antzekotasuna modelatzeko. Gure esperimentuek proposatutako ereduen baliagarritasuna erakusten dute, emaitza itxaropentsuak eskuratuz oinarrizko sistema sendo baten gainean. Era berean, gure lanak ekarpen garrantzitsuak egiten ditu errepresentazio bektorialen mapaketa elebidunei eta hitzen errepresentazio bektorialetan oinarritutako hitz-segiden antzekotasun neurriei dagokienean, itzulpen automatikoaz haratago balio propio bat dutenak semantika distribuzionalaren arloan.

Pose and Structure Recovery using Active Models

A new formulation of a pose refinement technique using ``active'' models is described. An error term derived from the detection of image derivatives close to an initial object hypothesis is linearised and solved by least squares. The method is particularly well suited to problems involving external geometrical constraints (such as the ground-plane constraint). We show that the method is able to recover both the pose of a rigid model, and the structure of a deformable model. We report an initial assessment of the performance and cost of pose and structure recovery using the active model in comparison with our previously reported ``passive'' model-based techniques in the context of traffic surveillance. The new method is more stable, and requires fewer iterations, especially when the number of free parameters increases, but shows somewhat poorer convergence.

Simulating SOC changes in 11 land use change chronosequences from the Brazilian Amazon with RothC and Century models

Land use and land cover changes in the Brazilian Amazon have major implications for regional and global carbon (C) cycling. Cattle pasture represents the largest single use (about 70%) of this once-forested land in most of the region. The main objective of this study was to evaluate the accuracy of the RothC and Century models at estimating soil organic C (SOC) changes under forest-to-pasture conditions in the Brazilian Amazon. We used data from 11 site-specific 'forest to pasture' chronosequences with the Century Ecosystem Model (Century 4.0) and the Rothamsted C Model (RothC 26.3). The models predicted that forest clearance and conversion to well managed pasture would cause an initial decline in soil C stocks (0-20 cm depth), followed in the majority of cases by a slow rise to levels exceeding those under native forest. One exception to this pattern was a chronosequence in Suia-Missu, which is under degraded pasture. In three other chronosequences the recovery of soil C under pasture appeared to be only to about the same level as under the previous forest. Statistical tests were applied to determine levels of agreement between simulated SOC stocks and observed stocks for all the sites within the 11 chronosequences. The models also provided reasonable estimates (coefficient of correlation = 0.8) of the microbial biomass C in the 0-10 cm soil layer for three chronosequences, when compared with available measured data. The Century model adequately predicted the magnitude and the overall trend in delta C-13 for the six chronosequences where measured 813 C data were available. This study gave independent tests of model performance, as no adjustments were made to the models to generate outputs. Our results suggest that modelling techniques can be successfully used for monitoring soil C stocks and changes, allowing both the identification of current patterns in the soil and the projection of future conditions. Results were used and discussed not only to evaluate soil C dynamics but also to indicate soil C sequestration opportunities for the Brazilian Amazon region. Moreover, modelling studies in these 'forest to pasture' systems have important applications, for example, the calculation of CO, emissions from land use change in national greenhouse gas inventories. (0 2007 Elsevier B.V. All rights reserved.

Diagnosis of variability and trends in a global precipitation dataset using a physically motivated statistical model

A physically motivated statistical model is used to diagnose variability and trends in wintertime ( October - March) Global Precipitation Climatology Project (GPCP) pentad (5-day mean) precipitation. Quasi-geostrophic theory suggests that extratropical precipitation amounts should depend multiplicatively on the pressure gradient, saturation specific humidity, and the meridional temperature gradient. This physical insight has been used to guide the development of a suitable statistical model for precipitation using a mixture of generalized linear models: a logistic model for the binary occurrence of precipitation and a Gamma distribution model for the wet day precipitation amount. The statistical model allows for the investigation of the role of each factor in determining variations and long-term trends. Saturation specific humidity q(s) has a generally negative effect on global precipitation occurrence and with the tropical wet pentad precipitation amount, but has a positive relationship with the pentad precipitation amount at mid- and high latitudes. The North Atlantic Oscillation, a proxy for the meridional temperature gradient, is also found to have a statistically significant positive effect on precipitation over much of the Atlantic region. Residual time trends in wet pentad precipitation are extremely sensitive to the choice of the wet pentad threshold because of increasing trends in low-amplitude precipitation pentads; too low a choice of threshold can lead to a spurious decreasing trend in wet pentad precipitation amounts. However, for not too small thresholds, it is found that the meridional temperature gradient is an important factor for explaining part of the long-term trend in Atlantic precipitation.