Automatic musical instrument recognition for multimedia indexing

Trabalho apresentado no âmbito do Mestrado em Engenharia Informática, como requisito parcial para obtenção do grau de Mestre em Engenharia Informática

Stratigraphic interpretation of Well-Log data of the Athabasca Oil Sands of Alberta Canada through Pattern recognition and Artificial Intelligence

Dissertation submitted in partial fulfillment of the requirements for the Degree of Master of Science in Geospatial Technologies.

Human activity recognition with accelerometry: novel time and frequency features

Human Activity Recognition systems require objective and reliable methods that can be used in the daily routine and must offer consistent results according with the performed activities. These systems are under development and offer objective and personalized support for several applications such as the healthcare area. This thesis aims to create a framework for human activities recognition based on accelerometry signals. Some new features and techniques inspired in the audio recognition methodology are introduced in this work, namely Log Scale Power Bandwidth and the Markov Models application. The Forward Feature Selection was adopted as the feature selection algorithm in order to improve the clustering performances and limit the computational demands. This method selects the most suitable set of features for activities recognition in accelerometry from a 423th dimensional feature vector. Several Machine Learning algorithms were applied to the used accelerometry databases – FCHA and PAMAP databases - and these showed promising results in activities recognition. The developed algorithm set constitutes a mighty contribution for the development of reliable evaluation methods of movement disorders for diagnosis and treatment applications.

Predicting market direction with hidden Markov models

This paper develops the model of Bicego, Grosso, and Otranto (2008) and applies Hidden Markov Models to predict market direction. The paper draws an analogy between financial markets and speech recognition, seeking inspiration from the latter to solve common issues in quantitative investing. Whereas previous works focus mostly on very complex modifications of the original hidden markov model algorithm, the current paper provides an innovative methodology by drawing inspiration from thoroughly tested, yet simple, speech recognition methodologies. By grouping returns into sequences, Hidden Markov Models can then predict market direction the same way they are used to identify phonemes in speech recognition. The model proves highly successful in identifying market direction but fails to consistently identify whether a trend is in place. All in all, the current paper seeks to bridge the gap between speech recognition and quantitative finance and, even though the model is not fully successful, several refinements are suggested and the room for improvement is significant.

Electronic tongue: a versatile tool for mineral and fruit-flavored waters recognition

Natural mineral waters (still), effervescent natural mineral waters (sparkling) and aromatized waters with fruit-flavors (still or sparkling) are an emerging market. In this work, the capability of a potentiometric electronic tongue, comprised with lipid polymeric membranes, to quantitatively estimate routinely quality physicochemical parameters (pH and conductivity) as well as to qualitatively classify water samples according to the type of water was evaluated. The study showed that a linear discriminant model, based on 21 sensors selected by the simulated annealing algorithm, could correctly classify 100 % of the water samples (leave-one out cross-validation). This potential was further demonstrated by applying a repeated K-fold cross-validation (guaranteeing that at least 15 % of independent samples were only used for internal-validation) for which 96 % of correct classifications were attained. The satisfactory recognition performance of the E-tongue could be attributed to the pH, conductivity, sugars and organic acids contents of the studied waters, which turned out in significant differences of sweetness perception indexes and total acid flavor. Moreover, the E-tongue combined with multivariate linear regression models, based on sub-sets of sensors selected by the simulated annealing algorithm, could accurately estimate waters pH (25 sensors: R 2 equal to 0.99 and 0.97 for leave-one-out or repeated K-folds cross-validation) and conductivity (23 sensors: R 2 equal to 0.997 and 0.99 for leave-one-out or repeated K-folds cross-validation). So, the overall satisfactory results achieved, allow envisaging a potential future application of electronic tongue devices for bottled water analysis and classification.

Dance movement patterns recognition

"Es tracta d'un projecte dividit en dues parts independents però complementàries, realitzades per autors diferents. Aquest document conté originàriament altre material i/o programari només consultable a la Biblioteca de Ciència i Tecnologia"

Object recognition for autonomous objects: comparison of two approaches

Report for the scientific sojourn at the Swiss Federal Institute of Technology Zurich, Switzerland, between September and December 2007. In order to make robots useful assistants for our everyday life, the ability to learn and recognize objects is of essential importance. However, object recognition in real scenes is one of the most challenging problems in computer vision, as it is necessary to deal with difficulties. Furthermore, in mobile robotics a new challenge is added to the list: computational complexity. In a dynamic world, information about the objects in the scene can become obsolete before it is ready to be used if the detection algorithm is not fast enough. Two recent object recognition techniques have achieved notable results: the constellation approach proposed by Lowe and the bag of words approach proposed by Nistér and Stewénius. The Lowe constellation approach is the one currently being used in the robot localization project of the COGNIRON project. This report is divided in two main sections. The first section is devoted to briefly review the currently used object recognition system, the Lowe approach, and bring to light the drawbacks found for object recognition in the context of indoor mobile robot navigation. Additionally the proposed improvements for the algorithm are described. In the second section the alternative bag of words method is reviewed, as well as several experiments conducted to evaluate its performance with our own object databases. Furthermore, some modifications to the original algorithm to make it suitable for object detection in unsegmented images are proposed.

Use of Hy's law and a new composite algorithm to predict acute liver failure in patients with drug-induced liver injury.

BACKGROUND & AIMS Hy's Law, which states that hepatocellular drug-induced liver injury (DILI) with jaundice indicates a serious reaction, is used widely to determine risk for acute liver failure (ALF). We aimed to optimize the definition of Hy's Law and to develop a model for predicting ALF in patients with DILI. METHODS We collected data from 771 patients with DILI (805 episodes) from the Spanish DILI registry, from April 1994 through August 2012. We analyzed data collected at DILI recognition and at the time of peak levels of alanine aminotransferase (ALT) and total bilirubin (TBL). RESULTS Of the 771 patients with DILI, 32 developed ALF. Hepatocellular injury, female sex, high levels of TBL, and a high ratio of aspartate aminotransferase (AST):ALT were independent risk factors for ALF. We compared 3 ways to use Hy's Law to predict which patients would develop ALF; all included TBL greater than 2-fold the upper limit of normal (×ULN) and either ALT level greater than 3 × ULN, a ratio (R) value (ALT × ULN/alkaline phosphatase × ULN) of 5 or greater, or a new ratio (nR) value (ALT or AST, whichever produced the highest ×ULN/ alkaline phosphatase × ULN value) of 5 or greater. At recognition of DILI, the R- and nR-based models identified patients who developed ALF with 67% and 63% specificity, respectively, whereas use of only ALT level identified them with 44% specificity. However, the level of ALT and the nR model each identified patients who developed ALF with 90% sensitivity, whereas the R criteria identified them with 83% sensitivity. An equal number of patients who did and did not develop ALF had alkaline phosphatase levels greater than 2 × ULN. An algorithm based on AST level greater than 17.3 × ULN, TBL greater than 6.6 × ULN, and AST:ALT greater than 1.5 identified patients who developed ALF with 82% specificity and 80% sensitivity. CONCLUSIONS When applied at DILI recognition, the nR criteria for Hy's Law provides the best balance of sensitivity and specificity whereas our new composite algorithm provides additional specificity in predicting the ultimate development of ALF.

OSIRISv1.2: a named entity recognition system for sequence variants of genes in biomedical literature

Background: Single Nucleotide Polymorphisms, among other type of sequence variants, constitute key elements in genetic epidemiology and pharmacogenomics. While sequence data about genetic variation is found at databases such as dbSNP, clues about the functional and phenotypic consequences of the variations are generally found in biomedical literature. The identification of the relevant documents and the extraction of the information from them are hampered by the large size of literature databases and the lack of widely accepted standard notation for biomedical entities. Thus, automatic systems for the identification of citations of allelic variants of genes in biomedical texts are required. Results: Our group has previously reported the development of OSIRIS, a system aimed at the retrieval of literature about allelic variants of genes http://ibi.imim.es/osirisform.html. Here we describe the development of a new version of OSIRIS (OSIRISv1.2, http://ibi.imim.es/OSIRISv1.2.html webcite) which incorporates a new entity recognition module and is built on top of a local mirror of the MEDLINE collection and HgenetInfoDB: a database that collects data on human gene sequence variations. The new entity recognition module is based on a pattern-based search algorithm for the identification of variation terms in the texts and their mapping to dbSNP identifiers. The performance of OSIRISv1.2 was evaluated on a manually annotated corpus, resulting in 99% precision, 82% recall, and an F-score of 0.89. As an example, the application of the system for collecting literature citations for the allelic variants of genes related to the diseases intracranial aneurysm and breast cancer is presented. Conclusion: OSIRISv1.2 can be used to link literature references to dbSNP database entries with high accuracy, and therefore is suitable for collecting current knowledge on gene sequence variations and supporting the functional annotation of variation databases. The application of OSIRISv1.2 in combination with controlled vocabularies like MeSH provides a way to identify associations of biomedical interest, such as those that relate SNPs with diseases.

Feature Extraction from Whole-Sky Ground-Based Images for Cloud-Type Recognition

Several features that can be extracted from digital images of the sky and that can be useful for cloud-type classification of such images are presented. Some features are statistical measurements of image texture, some are based on the Fourier transform of the image and, finally, others are computed from the image where cloudy pixels are distinguished from clear-sky pixels. The use of the most suitable features in an automatic classification algorithm is also shown and discussed. Both the features and the classifier are developed over images taken by two different camera devices, namely, a total sky imager (TSI) and a whole sky imager (WSC), which are placed in two different areas of the world (Toowoomba, Australia; and Girona, Spain, respectively). The performance of the classifier is assessed by comparing its image classification with an a priori classification carried out by visual inspection of more than 200 images from each camera. The index of agreement is 76% when five different sky conditions are considered: clear, low cumuliform clouds, stratiform clouds (overcast), cirriform clouds, and mottled clouds (altocumulus, cirrocumulus). Discussion on the future directions of this research is also presented, regarding both the use of other features and the use of other classification techniques

Nestmate recognition in the unicolonial ant Formica paralugubris

In unicolonial populations of ants, individuals can mix freely within large networks of nests that contain many queens. It has been proposed that the absence of aggression in unicolonial populations stems from a loss of nest mate recognition, but few studies have tested this hypothesis. We investigated patterns of aggression and nest mate recognition in the unicolonial wood ant, Formica paralugubris. Little aggression occurred, even between workers from nests separated by up to 5 km. However, when aggression took place, it was directed toward non-nest mates rather than nest mates. Trophallaxis (exchange of liquid food) occurred very frequently, and surprisingly, workers performed significantly more trophallaxis with non-nest mates than with nest mates (bias 2.4:1). Hence, workers are able to discriminate nest mates from non-nest mates. Higher rates of trophallaxis between non-nest mates may serve to homogenize the colony odor or may be an appeasement mechanism. Trophallaxis rate and aggression level were not correlated with geographical distance and did not differ within and between two populations separated by several kilometers. Hence, these populations do not represent differentiated supercolonies with clear-cut behavioral boundaries. Overall, the data demonstrate that unicoloniality can evolve despite well-developed nest mate recognition. Reduced levels of aggression might have been favored by the low rate of interactions with foreign workers, high cost of erroneously rejecting nest mates, and low cost of accepting foreign workers.

Efficient total variation algorithm for fetal brain MRI reconstruction.

Fetal MRI reconstruction aims at finding a high-resolution image given a small set of low-resolution images. It is usually modeled as an inverse problem where the regularization term plays a central role in the reconstruction quality. Literature has considered several regularization terms s.a. Dirichlet/Laplacian energy, Total Variation (TV)- based energies and more recently non-local means. Although TV energies are quite attractive because of their ability in edge preservation, standard explicit steepest gradient techniques have been applied to optimize fetal-based TV energies. The main contribution of this work lies in the introduction of a well-posed TV algorithm from the point of view of convex optimization. Specifically, our proposed TV optimization algorithm or fetal reconstruction is optimal w.r.t. the asymptotic and iterative convergence speeds O(1/n2) and O(1/√ε), while existing techniques are in O(1/n2) and O(1/√ε). We apply our algorithm to (1) clinical newborn data, considered as ground truth, and (2) clinical fetal acquisitions. Our algorithm compares favorably with the literature in terms of speed and accuracy.

The evolution of unicoloniality in ant societies : recognition, dispersal and population genetic structure

Introduction Societies of ants, bees, wasps and termites dominate many terrestrial ecosystems (Wilson 1971). Their evolutionary and ecological success is based upon the regulation of internal conflicts (e.g. Ratnieks et al. 2006), control of diseases (e.g. Schmid-Hempel 1998) and individual skills and collective intelligence in resource acquisition, nest building and defence (e.g. Camazine 2001). Individuals in social species can pass on their genes not only directly trough their own offspring, but also indirectly by favouring the reproduction of relatives. The inclusive fitness theory of Hamilton (1963; 1964) provides a powerful explanation for the evolution of reproductive altruism and cooperation in groups with related individuals. The same theory also led to the realization that insect societies are subject to internal conflicts over reproduction. Relatedness of less-than-one is not sufficient to eliminate all incentive for individual selfishness. This would indeed require a relatedness of one, as found among cells of an organism (Hardin 1968; Keller 1999). The challenge for evolutionary biology is to understand how groups can prevent or reduce the selfish exploitation of resources by group members, and how societies with low relatedness are maintained. In social insects the evolutionary shift from single- to multiple queens colonies modified the relatedness structure, the dispersal, and the mode of colony founding (e.g. (Crozier & Pamilo 1996). In ants, the most common, and presumably ancestral mode of reproduction is the emission of winged males and females, which found a new colony independently after mating and dispersal flights (Hölldobler & Wilson 1990). The alternative reproductive tactic for ant queens in multiple-queen colonies (polygyne) is to seek to be re-accepted in their natal colonies, where they may remain as additional reproductives or subsequently disperse on foot with part of the colony (budding) (Bourke & Franks 1995; Crozier & Pamilo 1996; Hölldobler & Wilson 1990). Such ant colonies can contain up to several hundred reproductive queens with an even more numerous workforce (Cherix 1980; Cherix 1983). As a consequence in polygynous ants the relatedness among nestmates is very low, and workers raise brood of queens to which they are only distantly related (Crozier & Pamilo 1996; Queller & Strassmann 1998). Therefore workers could increase their inclusive fitness by preferentially caring for their closest relatives and discriminate against less related or foreign individuals (Keller 1997; Queller & Strassmann 2002; Tarpy et al. 2004). However, the bulk of the evidence suggests that social insects do not behave nepotistically, probably because of the costs entailed by decreased colony efficiency or discrimination errors (Keller 1997). Recently, the consensus that nepotistic behaviour does not occur in insect colonies was challenged by a study in the ant Formica fusca (Hannonen & Sundström 2003b) showing that the reproductive share of queens more closely related to workers increases during brood development. However, this pattern can be explained either by nepotism with workers preferentially rearing the brood of more closely related queens or intrinsic differences in the viability of eggs laid by queens. In the first chapter, we designed an experiment to disentangle nepotism and differences in brood viability. We tested if workers prefer to rear their kin when given the choice between highly related and unrelated brood in the ant F. exsecta. We also looked for differences in egg viability among queens and simulated if such differences in egg viability may mistakenly lead to the conclusion that workers behave nepotistically. The acceptance of queens in polygnous ants raises the question whether the varying degree of relatedness affects their share in reproduction. In such colonies workers should favour nestmate queens over foreign queens. Numerous studies have investigated reproductive skew and partitioning of reproduction among queens (Bourke et al. 1997; Fournier et al. 2004; Fournier & Keller 2001; Hammond et al. 2006; Hannonen & Sundström 2003a; Heinze et al. 2001; Kümmerli & Keller 2007; Langer et al. 2004; Pamilo & Seppä 1994; Ross 1988; Ross 1993; Rüppell et al. 2002), yet almost no information is available on whether differences among queens in their relatedness to other colony members affects their share in reproduction. Such data are necessary to compare the relative reproductive success of dispersing and non-dispersing individuals. Moreover, information on whether there is a difference in reproductive success between resident and dispersing queens is also important for our understanding of the genetic structure of ant colonies and the dynamics of within group conflicts. In chapter two, we created single-queen colonies and then introduced a foreign queens originating from another colony kept under similar conditions in order to estimate the rate of queen acceptance into foreign established colonies, and to quantify the reproductive share of resident and introduced queens. An increasing number of studies have investigated the discrimination ability between ant workers (e.g. Holzer et al. 2006; Pedersen et al. 2006), but few have addressed the recognition and discrimination behaviour of workers towards reproductive individuals entering colonies (Bennett 1988; Brown et al. 2003; Evans 1996; Fortelius et al. 1993; Kikuchi et al. 2007; Rosengren & Pamilo 1986; Stuart et al. 1993; Sundström 1997; Vásquez & Silverman in press). These studies are important, because accepting new queens will generally have a large impact on colony kin structure and inclusive fitness of workers (Heinze & Keller 2000). In chapter three, we examined whether resident workers reject young foreign queens that enter into their nest. We introduced mated queens into their natal nest, a foreign-female producing nest, or a foreign male-producing nest and measured their survival. In addition, we also introduced young virgin and mated queens into their natal nest to examine whether the mating status of the queens influences their survival and acceptance by workers. On top of polgyny, some ant species have evolved an extraordinary social organization called 'unicoloniality' (Hölldobler & Wilson 1977; Pedersen et al. 2006). In unicolonial ants, intercolony borders are absent and workers and queens mix among the physically separated nests, such that nests form one large supercolony. Super-colonies can become very large, so that direct cooperative interactions are impossible between individuals of distant nests. Unicoloniality is an evolutionary paradox and a potential problem for kin selection theory because the mixing of queens and workers between nests leads to extremely low relatedness among nestmates (Bourke & Franks 1995; Crozier & Pamilo 1996; Keller 1995). A better understanding of the evolution and maintenance of unicoloniality requests detailed information on the discrimination behavior, dispersal, population structure, and the scale of competition. Cryptic genetic population structure may provide important information on the relevant scale to be considered when measuring relatedness and the role of kin selection. Theoretical studies have shown that relatedness should be measured at the level of the `economic neighborhood', which is the scale at which intraspecific competition generally takes place (Griffin & West 2002; Kelly 1994; Queller 1994; Taylor 1992). In chapter four, we conducted alarge-scale study to determine whether the unicolonial ant Formica paralugubris forms populations that are organised in discrete supercolonies or whether there is a continuous gradation in the level of aggression that may correlate with genetic isolation by distance and/or spatial distance between nests. In chapter five, we investigated the fine-scale population structure in three populations of F. paralugubris. We have developed mitochondria) markers, which together with the nuclear markers allowed us to detect cryptic genetic clusters of nests, to obtain more precise information on the genetic differentiation within populations, and to separate male and female gene flow. These new data provide important information on the scale to be considered when measuring relatedness in native unicolonial populations.

Place branding : a conceptual and theoretical framework

Within the context of rising competition between territories, identity has become the most important element of recognition, differentiation and commodification in the communicative process within which cities, regions and countries position themselves. Geographical spaces thus compete in terms of this identity, which is then subjected to fierce comparison and competition (Nogué, 1999; Anholt, 2007a). The territorial brand thus entails the reinvention of places through a process of brand construction (branding) based on the promotion of the individual and collective identities of geographical spaces; these identities, in turn, are imbued with the intangible factors associated with their respective territorial identities.

Typing less common ovarian tumors: A training tool based on a pattern-based algorithm applied to a set of 20 virtual slides.

Context: Ovarian tumors (OT) typing is a competency expected from pathologists, with significant clinical implications. OT however come in numerous different types, some rather rare, with the consequence of few opportunities for practice in some departments. Aim: Our aim was to design a tool for pathologists to train in less common OT typing. Method and Results: Representative slides of 20 less common OT were scanned (Nano Zoomer Digital Hamamatsu®) and the diagnostic algorithm proposed by Young and Scully applied to each case (Young RH and Scully RE, Seminars in Diagnostic Pathology 2001, 18: 161-235) to include: recognition of morphological pattern(s); shortlisting of differential diagnosis; proposition of relevant immunohistochemical markers. The next steps of this project will be: evaluation of the tool in several post-graduate training centers in Europe and Québec; improvement of its design based on evaluation results; diffusion to a larger public. Discussion: In clinical medicine, solving many cases is recognized as of utmost importance for a novice to become an expert. This project relies on the virtual slides technology to provide pathologists with a learning tool aimed at increasing their skills in OT typing. After due evaluation, this model might be extended to other uncommon tumors.