Non-negative matrix decomposition approaches to frequency domain analysis of music audio signals

On étudie l’application des algorithmes de décomposition matricielles tel que la Factorisation Matricielle Non-négative (FMN), aux représentations fréquentielles de signaux audio musicaux. Ces algorithmes, dirigés par une fonction d’erreur de reconstruction, apprennent un ensemble de fonctions de base et un ensemble de coef- ficients correspondants qui approximent le signal d’entrée. On compare l’utilisation de trois fonctions d’erreur de reconstruction quand la FMN est appliquée à des gammes monophoniques et harmonisées: moindre carré, divergence Kullback-Leibler, et une mesure de divergence dépendente de la phase, introduite récemment. Des nouvelles méthodes pour interpréter les décompositions résultantes sont présentées et sont comparées aux méthodes utilisées précédemment qui nécessitent des connaissances du domaine acoustique. Finalement, on analyse la capacité de généralisation des fonctions de bases apprises par rapport à trois paramètres musicaux: l’amplitude, la durée et le type d’instrument. Pour ce faire, on introduit deux algorithmes d’étiquetage des fonctions de bases qui performent mieux que l’approche précédente dans la majorité de nos tests, la tâche d’instrument avec audio monophonique étant la seule exception importante.

Advances in scaling deep learning algorithms

Les algorithmes d'apprentissage profond forment un nouvel ensemble de méthodes puissantes pour l'apprentissage automatique. L'idée est de combiner des couches de facteurs latents en hierarchies. Cela requiert souvent un coût computationel plus elevé et augmente aussi le nombre de paramètres du modèle. Ainsi, l'utilisation de ces méthodes sur des problèmes à plus grande échelle demande de réduire leur coût et aussi d'améliorer leur régularisation et leur optimization. Cette thèse adresse cette question sur ces trois perspectives. Nous étudions tout d'abord le problème de réduire le coût de certains algorithmes profonds. Nous proposons deux méthodes pour entrainer des machines de Boltzmann restreintes et des auto-encodeurs débruitants sur des distributions sparses à haute dimension. Ceci est important pour l'application de ces algorithmes pour le traitement de langues naturelles. Ces deux méthodes (Dauphin et al., 2011; Dauphin and Bengio, 2013) utilisent l'échantillonage par importance pour échantilloner l'objectif de ces modèles. Nous observons que cela réduit significativement le temps d'entrainement. L'accéleration atteint 2 ordres de magnitude sur plusieurs bancs d'essai. Deuxièmement, nous introduisont un puissant régularisateur pour les méthodes profondes. Les résultats expérimentaux démontrent qu'un bon régularisateur est crucial pour obtenir de bonnes performances avec des gros réseaux (Hinton et al., 2012). Dans Rifai et al. (2011), nous proposons un nouveau régularisateur qui combine l'apprentissage non-supervisé et la propagation de tangente (Simard et al., 1992). Cette méthode exploite des principes géometriques et permit au moment de la publication d'atteindre des résultats à l'état de l'art. Finalement, nous considérons le problème d'optimiser des surfaces non-convexes à haute dimensionalité comme celle des réseaux de neurones. Tradionellement, l'abondance de minimum locaux était considéré comme la principale difficulté dans ces problèmes. Dans Dauphin et al. (2014a) nous argumentons à partir de résultats en statistique physique, de la théorie des matrices aléatoires, de la théorie des réseaux de neurones et à partir de résultats expérimentaux qu'une difficulté plus profonde provient de la prolifération de points-selle. Dans ce papier nous proposons aussi une nouvelle méthode pour l'optimisation non-convexe.

Online Learning of Non-stationary Sequences

We consider an online learning scenario in which the learner can make predictions on the basis of a fixed set of experts. The performance of each expert may change over time in a manner unknown to the learner. We formulate a class of universal learning algorithms for this problem by expressing them as simple Bayesian algorithms operating on models analogous to Hidden Markov Models (HMMs). We derive a new performance bound for such algorithms which is considerably simpler than existing bounds. The bound provides the basis for learning the rate at which the identity of the optimal expert switches over time. We find an analytic expression for the a priori resolution at which we need to learn the rate parameter. We extend our scalar switching-rate result to models of the switching-rate that are governed by a matrix of parameters, i.e. arbitrary homogeneous HMMs. We apply and examine our algorithm in the context of the problem of energy management in wireless networks. We analyze the new results in the framework of Information Theory.

Video activity extraction and reporting with incremental unsupervised learning

The present work presents a new method for activity extraction and reporting from video based on the aggregation of fuzzy relations. Trajectory clustering is first employed mainly to discover the points of entry and exit of mobiles appearing in the scene. In a second step, proximity relations between resulting clusters of detected mobiles and contextual elements from the scene are modeled employing fuzzy relations. These can then be aggregated employing typical soft-computing algebra. A clustering algorithm based on the transitive closure calculation of the fuzzy relations allows building the structure of the scene and characterises the ongoing different activities of the scene. Discovered activity zones can be reported as activity maps with different granularities thanks to the analysis of the transitive closure matrix. Taking advantage of the soft relation properties, activity zones and related activities can be labeled in a more human-like language. We present results obtained on real videos corresponding to apron monitoring in the Toulouse airport in France.

Color Segmentation using LVQ-Learning Vector Quantization

This thesis aims to present a color segmentation approach for traffic sign recognition based on LVQ neural networks. The RGB images were converted into HSV color space, and segmented using LVQ depending on the hue and saturation values of each pixel in the HSV color space. LVQ neural network was used to segment red, blue and yellow colors on the road and traffic signs to detect and recognize them. LVQ was effectively applied to 536 sampled images taken from different countries in different conditions with 89% accuracy and the execution time of each image among 31 images was calculated in between 0.726sec to 0.844sec. The method was tested in different environmental conditions and LVQ showed its capacity to reasonably segment color despite remarkable illumination differences. The results showed high robustness.

Algoritmo Q-learning como estratégia de exploração e/ou explotação para metaheurísticas GRASP e algoritmo genético

Techniques of optimization known as metaheuristics have achieved success in the resolution of many problems classified as NP-Hard. These methods use non deterministic approaches that reach very good solutions which, however, don t guarantee the determination of the global optimum. Beyond the inherent difficulties related to the complexity that characterizes the optimization problems, the metaheuristics still face the dilemma of xploration/exploitation, which consists of choosing between a greedy search and a wider exploration of the solution space. A way to guide such algorithms during the searching of better solutions is supplying them with more knowledge of the problem through the use of a intelligent agent, able to recognize promising regions and also identify when they should diversify the direction of the search. This way, this work proposes the use of Reinforcement Learning technique - Q-learning Algorithm - as exploration/exploitation strategy for the metaheuristics GRASP (Greedy Randomized Adaptive Search Procedure) and Genetic Algorithm. The GRASP metaheuristic uses Q-learning instead of the traditional greedy-random algorithm in the construction phase. This replacement has the purpose of improving the quality of the initial solutions that are used in the local search phase of the GRASP, and also provides for the metaheuristic an adaptive memory mechanism that allows the reuse of good previous decisions and also avoids the repetition of bad decisions. In the Genetic Algorithm, the Q-learning algorithm was used to generate an initial population of high fitness, and after a determined number of generations, where the rate of diversity of the population is less than a certain limit L, it also was applied to supply one of the parents to be used in the genetic crossover operator. Another significant change in the hybrid genetic algorithm is the proposal of a mutually interactive cooperation process between the genetic operators and the Q-learning algorithm. In this interactive/cooperative process, the Q-learning algorithm receives an additional update in the matrix of Q-values based on the current best solution of the Genetic Algorithm. The computational experiments presented in this thesis compares the results obtained with the implementation of traditional versions of GRASP metaheuristic and Genetic Algorithm, with those obtained using the proposed hybrid methods. Both algorithms had been applied successfully to the symmetrical Traveling Salesman Problem, which was modeled as a Markov decision process

Unsupervised manifold learning using Reciprocal kNN Graphs in image re-ranking and rank aggregation tasks

Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES)

Image segmentation through combined methods: watershed transform, unsupervised distance learning and normalized cut

Research on image processing has shown that combining segmentation methods may lead to a solid approach to extract semantic information from different sort of images. Within this context, the Normalized Cut (NCut) is usually used as a final partitioning tool for graphs modeled in some chosen method. This work explores the Watershed Transform as a modeling tool, using different criteria of the hierarchical Watershed to convert an image into an adjacency graph. The Watershed is combined with an unsupervised distance learning step that redistributes the graph weights and redefines the Similarity matrix, before the final segmentation step using NCut. Adopting the Berkeley Segmentation Data Set and Benchmark as a background, our goal is to compare the results obtained for this method with previous work to validate its performance.

The Impact of Teaching Presence in Intensive Online Courses on Perceived Learning and Sense of Community: A Mixed Methods Study

This mixed methods concurrent triangulation design study was predicated upon two models that advocated a connection between teaching presence and perceived learning: the Community of Inquiry Model of Online Learning developed by Garrison, Anderson, and Archer (2000); and the Online Interaction Learning Model by Benbunan-Fich, Hiltz, and Harasim (2005). The objective was to learn how teaching presence impacted students’ perceptions of learning and sense of community in intensive online distance education courses developed and taught by instructors at a regional comprehensive university. In the quantitative phase online surveys collected relevant data from participating students (N = 397) and selected instructional faculty (N = 32) during the second week of a three-week Winter Term. Student information included: demographics such as age, gender, employment status, and distance from campus; perceptions of teaching presence; sense of community; perceived learning; course length; and course type. The students claimed having positive relationships between teaching presence, perceived learning, and sense of community. The instructors showed similar positive relationships with no significant differences when the student and instructor data were compared. The qualitative phase consisted of interviews with 12 instructors who had completed the online survey and replied to all of the open-response questions. The two phases were integrated using a matrix generation, and the analysis allowed for conclusions regarding teaching presence, perceived learning, and sense of community. The findings were equivocal with regard to satisfaction with course length and the relative importance of the teaching presence components. A model was provided depicting relationships between and among teaching presence components, perceived learning, and sense of community in intensive online courses.

Inhibition of matrix metalloproteinases and tumour necrosis factor alpha converting enzyme as adjuvant therapy in pneumococcal meningitis

Matrix metalloproteinases (MMPs) and tumour necrosis factor alpha (TNF-alpha) converting enzyme (TACE) contribute synergistically to the pathophysiology of bacterial meningitis. TACE proteolytically releases several cell-surface proteins, including the proinflammatory cytokine TNF-alpha and its receptors. TNF-alpha in turn stimulates cells to produce active MMPs, which facilitate leucocyte extravasation and brain oedema by degradation of extracellular matrix components. In the present time-course studies of pneumococcal meningitis in infant rats, MMP-8 and -9 were 100- to 1000-fold transcriptionally upregulated, both in CSF cells and in brain tissue. Concentrations of TNF-alpha and MMP-9 in CSF peaked 12 h after infection and were closely correlated. Treatment with BB-1101 (15 mg/kg subcutaneously, twice daily), a hydroxamic acid-based inhibitor of MMP and TACE, downregulated the CSF concentration of TNF-alpha and decreased the incidences of seizures and mortality. Therapy with BB-1101, together with antibiotics, attenuated neuronal necrosis in the cortex and apoptosis in the hippocampus when given as a pretreatment at the time of infection and also when administration was started 18 h after infection. Functionally, the neuroprotective effect of BB-1101 preserved learning performance of rats assessed 3 weeks after the disease had been cured. Thus, combined inhibition of MMP and TACE offers a novel therapeutic strategy to prevent brain injury and neurological sequelae in bacterial meningitis.

Capability Matrix : A Framework for Analyzing Capabilities in Value Chains

This paper develops a Capability Matrix for analyzing capabilities of developing country firms that participate in global and national value chains. This is a generic framework to capture firm-level knowledge accumulation in the context of global and local industrial constellations, by integrating key elements of the global value chain (GVC) and technological capabilities (TC) approaches. The framework can visually portray characteristics of firms’ capabilities, and highlight a relatively overlooked factor in the GVC approach: local firms’ endogenous learning efforts in varieties of relationship with lead firms.

Interocular transfer in perceptual learning of a pop-out discrimination task.

The specificity of the improvement in perceptual learning is often used to localize the neuronal changes underlying this type of adult plasticity. We investigated a visual texture discrimination task previously reported to be accomplished preattentively and for which learning-related changes were inferred to occur at a very early level of the visual processing stream. The stimulus was a matrix of lines from which a target popped out, due to an orientation difference between the three target lines and the background lines. The task was to report the global orientation of the target and was performed monocularly. The subjects' performance improved dramatically with training over the course of 2-3 weeks, after which we tested the specificity of the improvement for the eye trained. In all subjects tested, there was complete interocular transfer of the learning effect. The neuronal correlate of this learning are therefore most likely localized in a visual area where input from the two eyes has come together.

Incorporating curvature information into on-line learning

We analyse the dynamics of a number of second order on-line learning algorithms training multi-layer neural networks, using the methods of statistical mechanics. We first consider on-line Newton's method, which is known to provide optimal asymptotic performance. We determine the asymptotic generalization error decay for a soft committee machine, which is shown to compare favourably with the result for standard gradient descent. Matrix momentum provides a practical approximation to this method by allowing an efficient inversion of the Hessian. We consider an idealized matrix momentum algorithm which requires access to the Hessian and find close correspondence with the dynamics of on-line Newton's method. In practice, the Hessian will not be known on-line and we therefore consider matrix momentum using a single example approximation to the Hessian. In this case good asymptotic performance may still be achieved, but the algorithm is now sensitive to parameter choice because of noise in the Hessian estimate. On-line Newton's method is not appropriate during the transient learning phase, since a suboptimal unstable fixed point of the gradient descent dynamics becomes stable for this algorithm. A principled alternative is to use Amari's natural gradient learning algorithm and we show how this method provides a significant reduction in learning time when compared to gradient descent, while retaining the asymptotic performance of on-line Newton's method.

Natural gradient matrix momentum

Natural gradient learning is an efficient and principled method for improving on-line learning. In practical applications there will be an increased cost required in estimating and inverting the Fisher information matrix. We propose to use the matrix momentum algorithm in order to carry out efficient inversion and study the efficacy of a single step estimation of the Fisher information matrix. We analyse the proposed algorithm in a two-layer network, using a statistical mechanics framework which allows us to describe analytically the learning dynamics, and compare performance with true natural gradient learning and standard gradient descent.

The theory of on-line learning: a statistical physics approach

In this paper we review recent theoretical approaches for analysing the dynamics of on-line learning in multilayer neural networks using methods adopted from statistical physics. The analysis is based on monitoring a set of macroscopic variables from which the generalisation error can be calculated. A closed set of dynamical equations for the macroscopic variables is derived analytically and solved numerically. The theoretical framework is then employed for defining optimal learning parameters and for analysing the incorporation of second order information into the learning process using natural gradient descent and matrix-momentum based methods. We will also briefly explain an extension of the original framework for analysing the case where training examples are sampled with repetition.