Analysis of multi-scale systemic risk in Brazil's financial market

This work analyzes whether the relationship between risk and returns predicted by the Capital Asset Pricing Model (CAPM) is valid in the Brazilian stock market. The analysis is based on discrete wavelet decomposition on different time scales. This technique allows to analyze the relationship between different time horizons, since the short-term ones (2 to 4 days) up to the long-term ones (64 to 128 days). The results indicate that there is a negative or null relationship between systemic risk and returns for Brazil from 2004 to 2007. As the average excess return of a market portfolio in relation to a risk-free asset during that period was positive, it would be expected this relationship to be positive. That is, higher systematic risk should result in higher excess returns, which did not occur. Therefore, during that period, appropriate compensation for systemic risk was not observed in the Brazilian market. The scales that proved to be most significant to the risk-return relation were the first three, which corresponded to short-term time horizons. When treating differently, year-by-year, and consequently separating positive and negative premiums, some relevance is found, during some years, in the risk/return relation predicted by the CAPM. However, this pattern did not persist throughout the years. Therefore, there is not any evidence strong enough confirming that the asset pricing follows the model.

New insights into the role of microRNAs in pancreatic ß-cell maturation and plasticity

Le diabète est une maladie chronique caractérisée par une élévation du taux de sucre dans le sang aussi appelé « glycémie » reflétant un état pathologique. L'élévation de la glycémie au long cours a des répercussions délétères sur nombreux de nos tissus et organes d'où l'apparition de complications sévères chez les sujets diabétiques pouvant atteindre les yeux, les reins, le système nerveux, le système cardiovasculaire et les membres inférieurs. La carence en une hormone essentielle à notre organisme, l'insuline, est au coeur du développement de la maladie. L'insuline induit la captation du glucose circulant dans le sang en excès suite à une prise alimentaire riche en glucides et favorise son utilisation et éventuellement son stockage dans les tissus tels que le foie, le tissu adipeux et les muscles. Ainsi, l'insuline est vitale pour réguler et maintenir stable notre niveau de glycémie. Les cellules bêta du pancréas sont les seules entités de notre corps capables de produire de l'insuline et une perte de fonctionnalité associée à leur destruction ont été mises en cause dans le processus pathologique du diabète de type 2. Cependant la pleine fonctionnalité et la maturation des cellules bêta n'apparaissent qu'après la naissance lorsque le pancréas en développement a atteint sa masse adulte définitive. Enfin, une fois la masse des cellules bêta définitive établie, leur nombre et volume restent relativement constants au cours de la vie adulte chez un sujet sain. Néanmoins, au cours de périodes critiques les besoins en insuline sont augmentés tel qu'observé chez les femmes enceintes et les personnes obèses qui ont une perte de sensibilité à l'insuline qui se traduit par la nécessité de sécréter plus d'insuline afin de maintenir une glycémie normale. Dans l'hypothèse où la compensation n'a pas lieu ou n'est pas aboutie, le diabète se développe. Le processus de maturation postnatale ainsi que les événements compensatoires sont donc des étapes essentielles et de nombreuses questions sont encore non résolues concernant l'identification des mécanismes les régulant. Parmi les acteurs potentiels figurent de petites molécules d'ARN découvertes récemment appelées microARNs et qui ont été rapidement suggérées très prometteuses dans l'identification de nouvelles cibles thérapeutiques dans le cadre du diabète et d'autres pathologies. Les microARNs vont réguler l'expression de notre génome sans en modifier la séquence, phénomène également appelé épigénétique, ce qui résulte en des différences de comportement et de fonction cellulaires. Les microARNs sont donc susceptibles de jouer un rôle clé dans l'ensemble des processus biologiques et notre environnement associé à nos prédispositions génétiques peuvent grandement modifier leur niveau et donc leur action, qui à son tour se répercutera sur notre état physiologique. En effet nous avons identifié des changements de microARNs dans les cellules d'îlots pancréatiques de modèles animaux (rats et souris) associés à un état de résistance à l'insuline (grossesse et obésité). Par le biais d'expériences in vitro sur des cellules bêta extraites de rats et conservées en culture, nous avons pu analyser de plus près l'implication des microARNs dans la capacité des cellules bêta à sécréter de l'insuline mais aussi à se multiplier et à survivre au sein d'un environnement toxique. Ainsi, nous avons identifié des microARNs qui participent positivement à la compensation des cellules bêta, sous la direction d'hormones telles les estrogènes ou d'une hormone libérée par l'intestin au cours de la digestion (l'inerétine GLP1) et qui est largement utilisée comme agent thérapeutique dans la médication contre le diabète. Dans un second temps nous avons utilisé une stratégie similaire afin de déterminer le rôle de microARNs préalablement détectés comme étant changés au cours du développement postnatal des cellules bêta chez le rat. Cette étude a également mené à l'identification de microARNs participant à la maturation et à l'expansion de la masse des cellules bêta sous l'influence de la composition du régime alimentaire et des besoins en insuline adéquats qui en dépendent. Ces études apportent la vision de nouveaux mécanismes moléculaires impliquant les microARNs et démontrant leur importance pour le bon fonctionnement des cellules bêta et leur capacité d'adaptation à l'environnement. -- Les cellules bêta sont une composante des îlots pancréatiques de Langerhans et sont des cellules hautement différenciées qui ont l'unique capacité de sécréter de l'insuline sous l'influence des nutriments suite à une prise alimentaire. L'insuline facilite l'incorporation de glucose dans ses tissus cibles tels le foie, le tissu adipeux et les muscles. Bien que les besoins en insuline soient relativement constants au cours de la vie d'un individu sain, certaines conditions associées à un état de résistance à l'insuline, telles la grossesse ou l'obésité, requièrent une libération d'insuline majorée. En cas de résistance à l'insuline, une dysfonction des cellules bêta plus ou moins associée à leur mort cellulaire, conduisent à une sécrétion d'insuline insuffisante et au développement d'une hyperglycémie chronique, caractéristique du diabète de type 2. Jusqu'à présent, les mécanismes moléculaires sous- jacents à la compensation des cellules bêta ou encore menant à leur dysfonction restent peu connus. Découverts récemment, les petits ARNs non-codant appelés microARNs (miARNs), suscitent un intérêt grandissant de par leur potentiel thérapeutique pour la prise en charge et le traitement du diabète. Les miARNs sont de puissants régulateurs de l'expression génique qui lient directement le 3'UTR de leurs ARN messagers cibles afin d'inhiber leur traduction ou d'induire leur dégradation, ce qui leur permet de contrôler des fonctions biologiques multiples. Ainsi, nous avons pris pour hypothèse que les miARNs pourraient jouer un rôle essentiel en maintenant la fonction des cellules bêta et des processus compensatoires afin de prévenir le développement du diabète. Lors d'une première étude, une analyse transcriptomique a permis l'identification de miARNs différemment exprimés au sein d'îlots pancréatiques de rattes gestantes. Parmi eux, le miR-338-3p a démontré la capacité de promouvoir la prolifération et la survie des cellules bêta exposées à des acides gras saturés et des cytokines pro-inflammatoires, sans altérer leur propriété sécrétrice d'insuline. Nous avons également identifié deux hormones reconnues pour leurs propriétés bénéfiques pour la physiologie de la cellule bêta, l'estradiol et l'incrétine GLP1, qui régulent les niveaux du miR-338-3p. Ce miARN intègre parfaitement les voies de signalisation de ces deux hormones dépendantes de l'AMP cyclique, afin de contrôler l'expression de nombreux gènes conduisant à son action biologique. Dans un projet ultérieur, notre objectif était de déterminer la contribution de miARNs dans l'acquisition de l'identité fonctionnelle des cellules bêta en période postnatale. En effet, directement après la naissance les cellules bêta sont reconnues pour être encore immatures et incapables de sécréter de l'insuline spécifiquement en réponse à l'élévation de la glycémie. Au contraire, la réponse insulinique induite par les acides aminés ainsi que la biosynthèse d'insuline sont déjà fonctionnelles. Nos recherches ont permis de montrer que les changements de miARNs corrélés avec l'apparition du phénotype sécrétoire en réponse au glucose, sont régis par la composition nutritionnelle du régime alimentaire et des besoins en insuline qui en découlent. En parallèle, le taux de prolifération des cellules bêta est considérablement réduit. Les miARNs que nous avons étudiés coordonnent des changements d'expression de gènes clés impliqués dans l'acquisition de propriétés vitales de la cellule bêta et dans la maintenancé de son identité propre. Enfin, ces études ont permis de clairement démontrer l'importance des miARNs dans la régulation de la fonction des cellules bêta pancréatiques. -- Beta-cells are highly differentiated cells localized in the pancreatic islets and are characterized by the unique property of secreting insulin in response to nutrient stimulation after meal intake. Insulin is then in charge of facilitating glucose uptake by insulin target tissues such as liver, adipose tissue and muscles. Despite insulin needs stay more or less constant throughout life of healthy individuals, there are circumstances such as during pregnancy or obesity which are associated to insulin resistance, where insulin needs are increased. In this context, defects in beta-cell function, sometimes associated with beta-cell loss, may result in the release of inappropriate amounts of insulin leading to chronic hyperglycemia, properly defined as type 2 diabetes mellitus. So far, the mechanisms underlying beta- cell compensation as well as beta-cell failure remain to be established. The recently discovered small non-coding RNAs called microRNAs (miRNAs) are emerging as interesting therapeutic targets and are bringing new hope for the treatment of diabetes. miRNAs display a massive potential in regulating gene expression by directly binding to the 3'UTR of messenger RNAs and by inhibiting their translation and/or stability, enabling them to modify a wide range of biological functions. In view of this, we hypothesized that miRNAs may play an essential role in preserving the functional beta-cell mass and permitting to fight against beta-cell exhaustion and decompensation that can lead to diabetes development. In a first study, global profiling in pancreatic islets of pregnant rats, a model of insulin resistance, led to the identification of a set of differentially expressed miRNAs. Among them, miR-338- 3p was found to promote beta-cell proliferation and survival upon exposure of islet cells to pro- apoptotic stimuli such as saturated fatty acids or pro-inflammatory cytokines, without impairment in their capacity to release insulin. We also discovered that miR-338-3p changes are driven by two hormones, the estradiol and the incretin GLP1, both well known for their beneficial impact on beta- cell physiology. Consistently, we found that miR-338-3p integrates the cAMP-dependent signaling pathways regulated by these two hormones in order to control the expression of numerous genes and execute its biological functions. In a second project, we aimed at determining whether miRNAs contribute to the acquisition of beta-cell identity. Indeed, we confirmed that right after birth beta-cells are still immature and are unable to secrete insulin specifically in response to elevated concentrations of glucose. In contrast, amino acid-stimulated insulin release as well as insulin biosynthesis are already fully functional. In parallel, newborn beta-cells are proliferating intensively within the expanding pancreas. Interestingly, we demonstrated that the miRNA changes and the subsequent acquisition of glucose responsiveness is influenced by the diet composition and the resulting insulin needs. At the same time, beta-cell proliferation declines. The miRNAs that we have identified orchestrate expression changes of essential genes involved in the acquisition of specific beta-cell properties and in the maintenance of a mature beta-cell identity. Altogether, these studies clearly demonstrate that miRNAs play important roles in the regulation of beta-cell function.

Whose DNA is this? How relevant a question? (a note for forensic scientists).

This communication seeks to draw the attention of researchers and practitioners dealing with forensic DNA profiling analyses to the following question: is a scientist's report, offering support to a hypothesis according to which a particular individual is the source of DNA detected during the analysis of a stain, relevant from the point of view of a Court of Justice? This question relates to skeptical views previously voiced by commentators mainly in the judicial area, but is avoided by a large majority of forensic scientists. Notwithstanding, the pivotal role of this question has recently been evoked during the international conference "The hidden side of DNA profiles. Artifacts, errors and uncertain evidence" held in Rome (April 27th to 28th, 2012). Indeed, despite the fact that this conference brought together some of the world's leading forensic DNA specialists, it appeared clearly that a huge gap still exists between questions lawyers are actually interested in, and the answers that scientists deliver to Courts in written reports or during oral testimony. Participants in the justice system, namely lawyers and jurors on the one hand and forensic geneticists on the other, unfortunately talk considerably different languages. It thus is fundamental to address this issue of communication about results of forensic DNA analyses, and open a dialogue with practicing non-scientists at large who need to make meaningful use of scientific results to approach and help solve judicial cases. This paper intends to emphasize the actuality of this topic and suggest beneficial ways ahead towards a more reasoned use of forensic DNA in criminal proceedings.

Heterogeneous discounting in consumption-investment problems. Time consistent solutions

[cat] En aquest treball s'analitza un model estocàstic en temps continu en el que l'agent decisor descompta les utilitats instantànies i la funció final amb taxes de preferència temporal constants però diferents. En aquest context es poden modelitzar problemes en els quals, quan el temps s'acosta al moment final, la valoració de la funció final incrementa en comparació amb les utilitats instantànies. Aquest tipus d'asimetria no es pot descriure ni amb un descompte estàndard ni amb un variable. Per tal d'obtenir solucions consistents temporalment es deriva l'equació de programació dinàmica estocàstica, les solucions de la qual són equilibris Markovians. Per a aquest tipus de preferències temporals, s'estudia el model clàssic de consum i inversió (Merton, 1971) per a les funcions d'utilitat del tipus CRRA i CARA, comparant els equilibris Markovians amb les solucions inconsistents temporalment. Finalment es discuteix la introducció del temps final aleatori.

Rpe65 microarrays, from genes to phosphorylated proteins

Purpose:To describe a novel in silico method to gather and analyze data from high-throughput heterogeneous experimental procedures, i.e. gene and protein expression arrays. Methods:Each microarray is assigned to a database which handles common data (names, symbols, antibody codes, probe IDs, etc.). Links between informations are automatically generated from knowledge obtained in freely accessible databases (NCBI, Swissprot, etc). Requests can be made from any point of entry and the displayed result is fully customizable. Results:The initial database has been loaded with two sets of data: a first set of data originating from an Affymetrix-based retinal profiling performed in an RPE65 knock-out mouse model of Leber's congenital amaurosis. A second set of data generated from a Kinexus microarray experiment done on the retinas from the same mouse model has been added. Queries display wild type versus knock out expressions at several time points for both genes and proteins. Conclusions:This freely accessible database allows for easy consultation of data and facilitates data mining by integrating experimental data and biological pathways.

Issue of time in generally covariant theories and the Komar-Bergmann approach to observables in general relativity

Diffeomorphism-induced symmetry transformations and time evolution are distinct operations in generally covariant theories formulated in phase space. Time is not frozen. Diffeomorphism invariants are consequently not necessarily constants of the motion. Time-dependent invariants arise through the choice of an intrinsic time, or equivalently through the imposition of time-dependent gauge fixation conditions. One example of such a time-dependent gauge fixing is the Komar-Bergmann use of Weyl curvature scalars in general relativity. An analogous gauge fixing is also imposed for the relativistic free particle and the resulting complete set time-dependent invariants for this exactly solvable model are displayed. In contrast with the free particle case, we show that gauge invariants that are simultaneously constants of motion cannot exist in general relativity. They vary with intrinsic time.

Heterogeneous discounting in consumption-investment problems. Time consistent solutions

[cat] En aquest treball s'analitza un model estocàstic en temps continu en el que l'agent decisor descompta les utilitats instantànies i la funció final amb taxes de preferència temporal constants però diferents. En aquest context es poden modelitzar problemes en els quals, quan el temps s'acosta al moment final, la valoració de la funció final incrementa en comparació amb les utilitats instantànies. Aquest tipus d'asimetria no es pot descriure ni amb un descompte estàndard ni amb un variable. Per tal d'obtenir solucions consistents temporalment es deriva l'equació de programació dinàmica estocàstica, les solucions de la qual són equilibris Markovians. Per a aquest tipus de preferències temporals, s'estudia el model clàssic de consum i inversió (Merton, 1971) per a les funcions d'utilitat del tipus CRRA i CARA, comparant els equilibris Markovians amb les solucions inconsistents temporalment. Finalment es discuteix la introducció del temps final aleatori.

Robust accelerated failure time regression

Robust estimators for accelerated failure time models with asymmetric (or symmetric) error distribution and censored observations are proposed. It is assumed that the error model belongs to a log-location-scale family of distributions and that the mean response is the parameter of interest. Since scale is a main component of mean, scale is not treated as a nuisance parameter. A three steps procedure is proposed. In the first step, an initial high breakdown point S estimate is computed. In the second step, observations that are unlikely under the estimated model are rejected or down weighted. Finally, a weighted maximum likelihood estimate is computed. To define the estimates, functions of censored residuals are replaced by their estimated conditional expectation given that the response is larger than the observed censored value. The rejection rule in the second step is based on an adaptive cut-off that, asymptotically, does not reject any observation when the data are generat ed according to the model. Therefore, the final estimate attains full efficiency at the model, with respect to the maximum likelihood estimate, while maintaining the breakdown point of the initial estimator. Asymptotic results are provided. The new procedure is evaluated with the help of Monte Carlo simulations. Two examples with real data are discussed.

Auditory spatio-temporal brain dynamics and their consequences for multisensory interactions in humans.

Recent multisensory research has emphasized the occurrence of early, low-level interactions in humans. As such, it is proving increasingly necessary to also consider the kinds of information likely extracted from the unisensory signals that are available at the time and location of these interaction effects. This review addresses current evidence regarding how the spatio-temporal brain dynamics of auditory information processing likely curtails the information content of multisensory interactions observable in humans at a given latency and within a given brain region. First, we consider the time course of signal propagation as a limitation on when auditory information (of any kind) can impact the responsiveness of a given brain region. Next, we overview the dual pathway model for the treatment of auditory spatial and object information ranging from rudimentary to complex environmental stimuli. These dual pathways are considered an intrinsic feature of auditory information processing, which are not only partially distinct in their associated brain networks, but also (and perhaps more importantly) manifest only after several tens of milliseconds of cortical signal processing. This architecture of auditory functioning would thus pose a constraint on when and in which brain regions specific spatial and object information are available for multisensory interactions. We then separately consider evidence regarding mechanisms and dynamics of spatial and object processing with a particular emphasis on when discriminations along either dimension are likely performed by specific brain regions. We conclude by discussing open issues and directions for future research.

Detection and chemical profiling of medicine counterfeits by Raman spectroscopy and chemometrics

Raman spectroscopy combined with chemometrics has recently become a widespread technique for the analysis of pharmaceutical solid forms. The application presented in this paper is the investigation of counterfeit medicines. This increasingly serious issue involves networks that are an integral part of industrialized organized crime. Efficient analytical tools are consequently required to fight against it. Quick and reliable authentication means are needed to allow the deployment of measures from the company and the authorities. For this purpose a method in two steps has been implemented here. The first step enables the identification of pharmaceutical tablets and capsules and the detection of their counterfeits. A nonlinear classification method, the Support Vector Machines (SVM), is computed together with a correlation with the database and the detection of Active Pharmaceutical Ingredient (API) peaks in the suspect product. If a counterfeit is detected, the second step allows its chemical profiling among former counterfeits in a forensic intelligence perspective. For this second step a classification based on Principal Component Analysis (PCA) and correlation distance measurements is applied to the Raman spectra of the counterfeits.

Energiataseen hallinta ja optimointi metsäteollisuusintegraatissa

Diplomityö on tehty UPM-Kymmene Oyj, Kaukaan tehtailla Lappeenrannassa. Integroidussa metsäteollisuudessa energiantuotanto koostuu yleensä sähkön- ja lämmöntuotannosta. Kaukaan tehtailla prosessien lämmöntarve saadaan katettua kokonaisuudessaan omalla tuotannolla, kun taas kulutetusta sähköstä ainoastaan puolet on tuotettu itse. Loput sähköntarpeesta joudutaan ostamaan ulkopuolelta. Tutkimuksen pääpaino on ollut selvittää, miten kustannukset ovat riippuvaisia energiantuotannosta erilaisissa käyttöolosuhteissa. Työn tuloksena on luotu tietokonepohjainen laskentamalli, jonka avulla Kaukaan tehtaiden energiantuotantoa voidaan ohjata taloudellisesti optimaalisimmalla tavalla kulloinkin vallitsevassa käyttötilanteessa. Lisäksi tutkimuksessa on analysoitu tehdasintegraatin lämmönkulutuksen seurannan mahdollisuuksia lämmönsiirtoverkon nykyisten mittausten perusteella. Työssä on kerrottu yleisesti metsäteollisuuden energiankulutuksesta Suomessa. Lisäksi on esitetty arvioita energiankulutuksen kehityksestä tulevaisuudessa sekä keinoja energiatehokkuuden parantamiseksi. Kaukaan tehtailla lämmönkulutuksen seurantaan käytettävät mittausmenetelmät ja -laitteet on esitelty virtausmittausten osalta sekä arvioitu nykyisten mittausten luotettavuutta ja riittävyyttä kokonaisvaltaisen lämpötaseen hallintaan. Kaukaan tehtaiden energiantuotantojärjestelmästä on luotu termodynaaminen malli, johon energiantuotannosta aiheutuneiden kustannusten laskenta perustuu. Energiantuotannon optimoinnilla pyritään määrittelemään tietyn tarkasteluhetken käyttötilanteessa taloudellisesti optimaalisin kattiloiden ajojärjestys. Tarkastelu on rajattu lämmöntuotannon lisäämisen osalta maakaasun käytön lisäämiseen ja höyryturbiinien ohitukseen. Sähkön ja maakaasun hinnan sekä ympäristön lämpötilan vaihtelujen vaikutusta optimaaliseen ajojärjestykseen on havainnollistettu esimerkkien avulla.

Visualizing Business Model Evolution with the Business Model Canvas: Concept and Tool

The Business Model Canvas (BMC) assists in the design of companies' business models. As strategies evolve so too does the business model. Unfortunately, each BMC is a standalone representation. Thus, there is a need to be able to describe transformation from one version of a business model to the next as well as to visualize these operations. To address this issue, and to contribute to computer-assisted business model design, we propose a set of design principles for business model evolution. We also demonstrate a tool that can assist in the creation and navigation of business model versions in a visual and user-friendly way

Using appearance and context for outdoor scene object classification

We propose a probabilistic object classifier for outdoor scene analysis as a first step in solving the problem of scene context generation. The method begins with a top-down control, which uses the previously learned models (appearance and absolute location) to obtain an initial pixel-level classification. This information provides us the core of objects, which is used to acquire a more accurate object model. Therefore, their growing by specific active regions allows us to obtain an accurate recognition of known regions. Next, a stage of general segmentation provides the segmentation of unknown regions by a bottom-strategy. Finally, the last stage tries to perform a region fusion of known and unknown segmented objects. The result is both a segmentation of the image and a recognition of each segment as a given object class or as an unknown segmented object. Furthermore, experimental results are shown and evaluated to prove the validity of our proposal

A feature-based object-oriented expert system to model and support product design

In this paper a computer program to model and support product design is presented. The product is represented through a hierarchical structure that allows the user to navigate across the product’s components, and it aims at facilitating each step of the detail design process. A graphical interface was also developed, which shows visually to the user the contents of the product structure. Features are used as building blocks for the parts that compose the product, and object-oriented methodology was used as a means to implement the product structure. Finally, an expert system was also implemented, whose knowledge base rules help the user design a product that meets design and manufacturing requirements.

Generative Part-Based Gabor Object Detector

Object detection is a fundamental task of computer vision that is utilized as a core part in a number of industrial and scientific applications, for example, in robotics, where objects need to be correctly detected and localized prior to being grasped and manipulated. Existing object detectors vary in (i) the amount of supervision they need for training, (ii) the type of a learning method adopted (generative or discriminative) and (iii) the amount of spatial information used in the object model (model-free, using no spatial information in the object model, or model-based, with the explicit spatial model of an object). Although some existing methods report good performance in the detection of certain objects, the results tend to be application specific and no universal method has been found that clearly outperforms all others in all areas. This work proposes a novel generative part-based object detector. The generative learning procedure of the developed method allows learning from positive examples only. The detector is based on finding semantically meaningful parts of the object (i.e. a part detector) that can provide additional information to object location, for example, pose. The object class model, i.e. the appearance of the object parts and their spatial variance, constellation, is explicitly modelled in a fully probabilistic manner. The appearance is based on bio-inspired complex-valued Gabor features that are transformed to part probabilities by an unsupervised Gaussian Mixture Model (GMM). The proposed novel randomized GMM enables learning from only a few training examples. The probabilistic spatial model of the part configurations is constructed with a mixture of 2D Gaussians. The appearance of the parts of the object is learned in an object canonical space that removes geometric variations from the part appearance model. Robustness to pose variations is achieved by object pose quantization, which is more efficient than previously used scale and orientation shifts in the Gabor feature space. Performance of the resulting generative object detector is characterized by high recall with low precision, i.e. the generative detector produces large number of false positive detections. Thus a discriminative classifier is used to prune false positive candidate detections produced by the generative detector improving its precision while keeping high recall. Using only a small number of positive examples, the developed object detector performs comparably to state-of-the-art discriminative methods.