Authentification biométrique par dynamique de frappe pour évaluation à distance utilisant SVM à une classe

La machine à vecteurs de support à une classe est un algorithme non-supervisé qui est capable d’apprendre une fonction de décision à partir de données d’une seule classe pour la détection d’anomalie. Avec les données d’entraînement d’une seule classe, elle peut identifier si une nouvelle donnée est similaire à l’ensemble d’entraînement. Dans ce mémoire, nous nous intéressons à la reconnaissance de forme de dynamique de frappe par la machine à vecteurs de support à une classe, pour l’authentification d’étudiants dans un système d’évaluation sommative à distance à l’Université Laval. Comme chaque étudiant à l’Université Laval possède un identifiant court, unique qu’il utilise pour tout accès sécurisé aux ressources informatiques, nous avons choisi cette chaîne de caractères comme support à la saisie de dynamique de frappe d’utilisateur pour construire notre propre base de données. Après avoir entraîné un modèle pour chaque étudiant avec ses données de dynamique de frappe, on veut pouvoir l’identifier et éventuellement détecter des imposteurs. Trois méthodes pour la classification ont été testées et discutées. Ainsi, nous avons pu constater les faiblesses de chaque méthode dans ce système. L’évaluation des taux de reconnaissance a permis de mettre en évidence leur dépendance au nombre de signatures ainsi qu’au nombre de caractères utilisés pour construire les signatures. Enfin, nous avons montré qu’il existe des corrélations entre le taux de reconnaissance et la dispersion dans les distributions des caractéristiques des signatures de dynamique de frappe.

How well are Australian children doing in the first year of school : a demographic analysis

Background: The transition to school is a sensitive period for children in relation to school success. In the early school years, children need to develop positive attitudes to school and have experiences that promote academic, behavioural and social competence. When children begin school there are higher expectations of responsibility and independence and in the year one class, there are more explicit academic goals for literacy and numeracy and more formal instruction. Most importantly, children’s early attitudes to learning and learning styles have an impact on later educational outcomes. Method: Data were drawn from The Longitudinal Study of Australian Children (LSAC). LSAC is a cross-sequential cohort study funded by the Australian Government. In these analyses, Wave 2 (2006) data for 2499 children in the Kindergarten Cohort were used. Children, at Wave 2, were in the first year of formal school. They had a mean age of 6.9 years (SD= 0.26). Measures included a 6-item measure of Approaches to Learning (task persistence, independence) and the Academic Rating Scales for language and literacy and mathematical thinking. Teachers rated their relationships with children on the short form of the STRS. Results: Girls were rated by their teachers as doing better than boys on Language and literacy, Approaches to learning; and they had a better relationship with their teacher. Children from an Aboriginal or Torres Strait Island (ATSI) background were rated as doing less well on Language and Literacy and Mathematical thinking and on their Approaches to learning. Children from high Socio Economic Position families are doing better on teacher rated Language and Literacy, Mathematical thinking, Approaches to learning and they had a better relationship with their teacher. Conclusions: Findings highlight the importance of key demographic variables in understanding children’s early school success.

Provoking critical awareness and intersubjectivity through "transformative storytelling"

Theories that inform pedagogical practices have positioned young children as innocent, pre-political and egocentric. This paper draws from an action research study that investigates the impact of “transformative storytelling”, where stories purposefully crafted to counter metanarratives, revealed the impact of human greed with one class of children aged five to six years of age. Derrida’s notion of “cinders” provided a concept for investigating the traces or imprints the language of story left behind, amidst the children’s comments and actions, enabling the possibilities of the history of these “cinders” (that is what informed these comments and actions) to be noticed. Readings of some of the children’s responses suggest that children aged five and six years can engage in political discourse through the provocation of “transformative storytelling”, and that their engagement demonstrated the consideration of others through critical awareness and intersubjectivity. These early readings raise questions regarding curriculum content and pedagogical practices in early years education and the validity of ongoing educational goals that incorporate critical awareness and intersubjectivity to equip students with communitarian strategies to counter the individualistic outlook of neoliberalist societies.

How well are Australian children doing in the first year of school?

Background: The transition to school is a sensitive period for children in relation to school success. In the early school years, children need to develop positive attitudes to school and have experiences that promote academic, behavioural and social competence. When children begin school there are higher expectations of responsibility and independence and in the year one class, there are more explicit academic goals for literacy and numeracy and more formal instruction. Most importantly, children’s early attitudes to learning and learning styles have an impact on later educational outcomes. Method: Data were drawn from The Longitudinal Study of Australian Children (LSAC). LSAC is a cross-sequential cohort study funded by the Australian Government. In these analyses, Wave 2 (2006) data for 2499 children in the Kindergarten Cohort were used. Children, at Wave 2, were in the first year of formal school. They had a mean age of 6.9 years (SD= 0.26). Measures included a 6-item measure of Approaches to Learning (task persistence, independence) and the Academic Rating Scales for language and literacy and mathematical thinking. Teachers rated their relationships with children on the short form of the STRS. Results: Girls were rated by their teachers as doing better than boys on Language and literacy, Approaches to learning; and they had a better relationship with their teacher. Children from an Aboriginal or Torres Strait Island (ATSI) background were rated as doing less well on Language and Literacy and Mathematical thinking and on their Approaches to learning. Children from high Socio Economic Position families are doing better on teacher rated Language and Literacy, Mathematical thinking, Approaches to learning and they had a better relationship with their teacher. Conclusions: Findings highlight the importance of key demographic variables in understanding children’s early school success.

An FPGA-based approach to multi-objective evolutionary algorithm for multi-disciplinary design optimisation

This paper investigates the field programmable gate array (FPGA) approach for multi-objective and multi-disciplinary design optimisation (MDO) problems. One class of optimisation method that has been well-studied and established for large and complex problems, such as those inherited in MDO, is multi-objective evolutionary algorithms (MOEAs). The MOEA, nondominated sorting genetic algorithm II (NSGA-II), is hardware implemented on an FPGA chip. The NSGA-II on FPGA application to multi-objective test problem suites has verified the designed implementation effectiveness. Results show that NSGA-II on FPGA is three orders of magnitude better than the PC based counterpart.

Estimating the probability of an invalid match for matching using the rank transform

A fundamental problem faced by stereo matching algorithms is the matching or correspondence problem. A wide range of algorithms have been proposed for the correspondence problem. For all matching algorithms, it would be useful to be able to compute a measure of the probability of correctness, or reliability of a match. This paper focuses in particular on one class for matching algorithms, which are based on the rank transform. The interest in these algorithms for stereo matching stems from their invariance to radiometric distortion, and their amenability to fast hardware implementation. This work differs from previous work in that it derives, from first principles, an expression for the probability of a correct match. This method was based on an enumeration of all possible symbols for matching. The theoretical results for disparity error prediction, obtained using this method, were found to agree well with experimental results. However, disadvantages of the technique developed in this chapter are that it is not easily applicable to real images, and also that it is too computationally expensive for practical window sizes. Nevertheless, the exercise provides an interesting and novel analysis of match reliability.

Bayesian indirect inference using a parametric auxiliary model

Indirect inference (II) is a methodology for estimating the parameters of an intractable (generative) model on the basis of an alternative parametric (auxiliary) model that is both analytically and computationally easier to deal with. Such an approach has been well explored in the classical literature but has received substantially less attention in the Bayesian paradigm. The purpose of this paper is to compare and contrast a collection of what we call parametric Bayesian indirect inference (pBII) methods. One class of pBII methods uses approximate Bayesian computation (referred to here as ABC II) where the summary statistic is formed on the basis of the auxiliary model, using ideas from II. Another approach proposed in the literature, referred to here as parametric Bayesian indirect likelihood (pBIL), we show to be a fundamentally different approach to ABC II. We devise new theoretical results for pBIL to give extra insights into its behaviour and also its differences with ABC II. Furthermore, we examine in more detail the assumptions required to use each pBII method. The results, insights and comparisons developed in this paper are illustrated on simple examples and two other substantive applications. The first of the substantive examples involves performing inference for complex quantile distributions based on simulated data while the second is for estimating the parameters of a trivariate stochastic process describing the evolution of macroparasites within a host based on real data. We create a novel framework called Bayesian indirect likelihood (BIL) which encompasses pBII as well as general ABC methods so that the connections between the methods can be established.

Violence in and around entertainment districts : a longitudinal analysis of the impact of late-night lockout legislation

Violence in entertainment districts is a major problem across urban landscapes throughout the world. Research shows that licensed premises are the third most common location for homicides and serious assaults, accounting for one in ten fatal and nonfatal assaults. One class of interventions that aims to reduce violence in entertainment districts involves the use of civil remedies: a group of strategies that use civil or regulatory measures as legal “levers” to reduce problem behavior. One specific civil remedy used to reduce problematic behavior in entertainment districts involves manipulation of licensed premise trading hours. This article uses generalized linear models to analyze the impact of lockout legislation on recorded violent offences in two entertainment districts in the Australian state of Queensland. Our research shows that 3 a.m. lockout legislation led to a direct and significant reduction in the number of violent incidents inside licensed premises. Indeed, the lockouts cut the level of violent crime inside licensed premises by half. Despite these impressive results for the control of violence inside licensed premises, we found no evidence that the lockout had any impact on violence on streets and footpaths outside licensed premises that were the site for more than 80 percent of entertainment district violence. Overall, however, our analysis suggests that lockouts are an important mechanism that helps to control the level of violence inside licensed premises but that finely grained contextual responses to alcohol-related problems are needed rather than one-size-fits-all solutions.

Application of anomaly technique in wind turbine bearing fault detection

Bearing faults are the most common cause of wind turbine failures. Unavailability and maintenance cost of wind turbines are becoming critically important, with their fast growing in electric networks. Early fault detection can reduce outage time and costs. This paper proposes Anomaly Detection (AD) machine learning algorithms for fault diagnosis of wind turbine bearings. The application of this method on a real data set was conducted and is presented in this paper. For validation and comparison purposes, a set of baseline results are produced using the popular one-class SVM methods to examine the ability of the proposed technique in detecting incipient faults.

Relative peripheral hyperopia does not predict development and progression of myopia in children

Purpose To test the hypothesis that relative peripheral hyperopia predicts development and progression of myopia. Methods Refraction along the horizontal visual field was measured under cycloplegia at visual field angles of 0°, ±15°, and ±30° at baseline, 1 and 2 years in over 1700 initially 7-year-old Chinese children, and at baseline and 1 year in over 1000 initially 14-year olds. One refraction classification for central refraction was “nonmyopia, myopia” (nM, M), consisting of nM greater than −0.50 diopters (D; spherical equivalent) and M less than or equal to −0.50 D. A second classification was “hyperopia, emmetropia, low myopia, and moderate/high myopia” (H, E, LM, MM) with H greater than or equal to +1.00 D, E, −0.49 to +0.99 D, LM, −2.99 to −0.50 D, and MM less than or equal to −3.00 D. Subclassifications were made on the basis of development and progression of myopia over the 2 years. Changes in central refraction over time were determined for different groups, and relative peripheral refraction over time was compared between different subgroups. Results Simple linear regression of central refraction as a function of relative peripheral refraction did not predict myopia progression as relative peripheral refraction became more hyperopic: relative peripheral hyperopia and relative peripheral myopia predicted significant myopia progression for 0% and 35% of group/visual field angle combinations, respectively. Subgroups who developed myopia did not have more initial relative peripheral hyperopia than subgroups who did not develop myopia. Conclusions Relative peripheral hyperopia does not predict development nor progression of myopia in children. This calls into question the efficacy of treatments that aim to slow progression of myopia in children by “treating” relative peripheral hyperopia.

Microbe-induced activation of inflammatory cytokine response in human cells

Human body is in continuous contact with microbes. Although many microbes are harmless or beneficial for humans, pathogenic microbes possess a threat to wellbeing. Antimicrobial protection is provided by the immune system, which can be functionally divided into two parts, namely innate and adaptive immunity. The key players of the innate immunity are phagocytic white blood cells such as neutrophils, monocytes, macrophages and dendritic cells (DCs), which constantly monitor the blood and peripheral tissues. These cells are armed for rapid activation upon microbial contact since they express a variety of microbe-recognizing receptors. Macrophages and DCs also act as antigen presenting cells (APCs) and play an important role in the development of adaptive immunity. The development of adaptive immunity requires intimate cooperation between APCs and T lymphocytes and results in microbe-specific immune responses. Moreover, adaptive immunity generates immunological memory, which rapidly and efficiently protects the host from reinfection. Properly functioning immune system requires efficient communication between cells. Cytokines are proteins, which mediate intercellular communication together with direct cell-cell contacts. Immune cells produce inflammatory cytokines rapidly following microbial contact. Inflammatory cytokines modulate the development of local immune response by binding to cell surface receptors, which results in the activation of intracellular signalling and modulates target cell gene expression. One class of inflammatory cytokines chemokines has a major role in regulating cellular traffic. Locally produced inflammatory chemokines guide the recruitment of effector cells to the site of inflammation during microbial infection. In this study two key questions were addressed. First, the ability of pathogenic and non-pathogenic Gram-positive bacteria to activate inflammatory cytokine and chemokine production in different human APCs was compared. In these studies macrophages and DCs were stimulated with pathogenic Steptococcus pyogenes or non-pathogenic Lactobacillus rhamnosus. The second aim of this thesis work was to analyze the role of pro-inflammatory cytokines in the regulation of microbe-induced chemokine production. In these studies bacteria-stimulated macrophages and influenza A virus-infected lung epithelial cells were used as model systems. The results of this study show that although macrophages and DCs share several common antimicrobial functions, these cells have significantly distinct responses against pathogenic and non-pathogenic Gram-positive bacteria. Macrophages were activated in a nearly similar fashion by pathogenic S. pyogenes and non-pathogenic L. rhamnosus. Both bacteria induced the production of similar core set of inflammatory chemokines consisting of several CC-class chemokines and CXCL8. These chemokines attract monocytes, neutrophils, dendritic cells and T cells. Thus, the results suggest that bacteria-activated macrophages efficiently recruit other effector cells to the site of inflammation. Moreover, macrophages seem to be activated by all bacteria irrespective of their pathogenicity. DCs, in contrast, were efficiently activated only by pathogenic S. pyogenes, which induced DC maturation and production of several inflammatory cytokines and chemokines. In contrast, L. rhamnosus-stimulated DCs matured only partially and, most importantly, these cells did not produce inflammatory cytokines or chemokines. L. rhamnosus-stimulated DCs had a phenotype of "semi-mature" DCs and this type of DCs have been suggested to enhance tolerogenic adaptive immune responses. Since DCs have an essential role in the development of adaptive immune response the results suggest that, in contrast to macrophages, DCs may be able to discriminate between pathogenic and non-pathogenic bacteria and thus mount appropriate inflammatory or tolerogenic adaptive immune response depending on the microbe in question. The results of this study also show that pro-inflammatory cytokines can contribute to microbe-induced chemokine production at multiple levels. S. pyogenes-induced type I interferon (IFN) was found to enhance the production of certain inflammatory chemokines in macrophages during bacterial stimulation. Thus, bacteria-induced chemokine production is regulated by direct (microbe-induced) and indirect (pro-inflammatory cytokine-induced) mechanisms during inflammation. In epithelial cells IFN- and tumor necrosis factor- (TNF-) were found to enhance the expression of PRRs and components of cellular signal transduction machinery. Pre-treatment of epithelial cells with these cytokines prior to virus infection resulted in markedly enhanced chemokine response compared to untreated cells. In conclusion, the results obtained from this study show that pro-inflammatory cytokines can enhance microbe-induced chemokine production during microbial infection by providing a positive feedback loop. In addition, pro-inflammatory cytokines can render normally low-responding cells to high chemokine producers via enhancement of microbial detection and signal transduction.

R1SVM: A randomised nonlinear approach to large-scale anomaly detection

The problem of unsupervised anomaly detection arises in a wide variety of practical applications. While one-class support vector machines have demonstrated their effectiveness as an anomaly detection technique, their ability to model large datasets is limited due to their memory and time complexity for training. To address this issue for supervised learning of kernel machines, there has been growing interest in random projection methods as an alternative to the computationally expensive problems of kernel matrix construction and sup-port vector optimisation. In this paper we leverage the theory of nonlinear random projections and propose the Randomised One-class SVM (R1SVM), which is an efficient and scalable anomaly detection technique that can be trained on large-scale datasets. Our empirical analysis on several real-life and synthetic datasets shows that our randomised 1SVM algorithm achieves comparable or better accuracy to deep auto encoder and traditional kernelised approaches for anomaly detection, while being approximately 100 times faster in training and testing.

An improved Monte Carlo scheme for simulation of synthesis of nanoparticles in reverse micelles

An improved Monte Carlo technique is presented in this work to simulate nanoparticle formation through a micellar route. The technique builds on the simulation technique proposed by Bandyopadhyaya et al. (Langmuir 2000, 16, 7139) which is general and rigorous but at the same time very computation intensive, so much so that nanoparticle formation in low occupancy systems cannot be simulated in reasonable time. In view of this, several strategies, rationalized by simple mathematical analyses, are proposed to accelerate Monte Carlo simulations. These are elimination of infructuous events, removal of excess reactant postreaction, and use of smaller micelle population a large number of times. Infructuous events include collision of an empty micelle with another empty one or with another one containing only one molecule or only a solid particle. These strategies are incorporated in a new simulation technique which divides the entire micelle population in four classes and shifts micelles from one class to other as the simulation proceeds. The simulation results, throughly tested using chi-square and other tests, show that the predictions of the improved technique remain unchanged, but with more than an order of magnitude decrease in computational effort for some of the simulations reported in the literature. A post priori validation scheme for the correctness of the simulation results has been utilized to propose a new simulation strategy to arrive at converged simulation results with near minimum computational effort.

The Lovasz θ function, SVMs and finding large dense subgraphs

The Lovasz θ function of a graph, is a fundamental tool in combinatorial optimization and approximation algorithms. Computing θ involves solving a SDP and is extremely expensive even for moderately sized graphs. In this paper we establish that the Lovasz θ function is equivalent to a kernel learning problem related to one class SVM. This interesting connection opens up many opportunities bridging graph theoretic algorithms and machine learning. We show that there exist graphs, which we call SVM−θ graphs, on which the Lovasz θ function can be approximated well by a one-class SVM. This leads to a novel use of SVM techniques to solve algorithmic problems in large graphs e.g. identifying a planted clique of size Θ(n√) in a random graph G(n,12). A classic approach for this problem involves computing the θ function, however it is not scalable due to SDP computation. We show that the random graph with a planted clique is an example of SVM−θ graph, and as a consequence a SVM based approach easily identifies the clique in large graphs and is competitive with the state-of-the-art. Further, we introduce the notion of a ''common orthogonal labeling'' which extends the notion of a ''orthogonal labelling of a single graph (used in defining the θ function) to multiple graphs. The problem of finding the optimal common orthogonal labelling is cast as a Multiple Kernel Learning problem and is used to identify a large common dense region in multiple graphs. The proposed algorithm achieves an order of magnitude scalability compared to the state of the art.

Prediction of Queue Waiting Times for Metascheduling on Parallel Batch Systems

Prediction of queue waiting times of jobs submitted to production parallel batch systems is important to provide overall estimates to users and can also help meta-schedulers make scheduling decisions. In this work, we have developed a framework for predicting ranges of queue waiting times for jobs by employing multi-class classification of similar jobs in history. Our hierarchical prediction strategy first predicts the point wait time of a job using dynamic k-Nearest Neighbor (kNN) method. It then performs a multi-class classification using Support Vector Machines (SVMs) among all the classes of the jobs. The probabilities given by the SVM for the class predicted using k-NN and its neighboring classes are used to provide a set of ranges of predicted wait times with probabilities. We have used these predictions and probabilities in a meta-scheduling strategy that distributes jobs to different queues/sites in a multi-queue/grid environment for minimizing wait times of the jobs. Experiments with different production supercomputer job traces show that our prediction strategies can give correct predictions for about 77-87% of the jobs, and also result in about 12% improved accuracy when compared to the next best existing method. Experiments with our meta-scheduling strategy using different production and synthetic job traces for various system sizes, partitioning schemes and different workloads, show that the meta-scheduling strategy gives much improved performance when compared to existing scheduling policies by reducing the overall average queue waiting times of the jobs by about 47%.