Evaluating retraining rules for semi-supervised learning in neural network based cursive word recognition

Training a system to recognize handwritten words is a task that requires a large amount of data with their correct transcription. However, the creation of such a training set, including the generation of the ground truth, is tedious and costly. One way of reducing the high cost of labeled training data acquisition is to exploit unlabeled data, which can be gathered easily. Making use of both labeled and unlabeled data is known as semi-supervised learning. One of the most general versions of semi-supervised learning is self-training, where a recognizer iteratively retrains itself on its own output on new, unlabeled data. In this paper we propose to apply semi-supervised learning, and in particular self-training, to the problem of cursive, handwritten word recognition. The special focus of the paper is on retraining rules that define what data are actually being used in the retraining phase. In a series of experiments it is shown that the performance of a neural network based recognizer can be significantly improved through the use of unlabeled data and self-training if appropriate retraining rules are applied.

Fully automatic X-ray image segmentation via joint estimation of image displacements.

We propose a new method for fully-automatic landmark detection and shape segmentation in X-ray images. Our algorithm works by estimating the displacements from image patches to the (unknown) landmark positions and then integrating them via voting. The fundamental contribution is that, we jointly estimate the displacements from all patches to multiple landmarks together, by considering not only the training data but also geometric constraints on the test image. The various constraints constitute a convex objective function that can be solved efficiently. Validated on three challenging datasets, our method achieves high accuracy in landmark detection, and, combined with statistical shape model, gives a better performance in shape segmentation compared to the state-of-the-art methods.

Sparse representation based action and gesture recognition

In this paper we present a solution to the problem of action and gesture recognition using sparse representations. The dictionary is modelled as a simple concatenation of features computed for each action or gesture class from the training data, and test data is classified by finding sparse representation of the test video features over this dictionary. Our method does not impose any explicit training procedure on the dictionary. We experiment our model with two kinds of features, by projecting (i) Gait Energy Images (GEIs) and (ii) Motion-descriptors, to a lower dimension using Random projection. Experiments have shown 100% recognition rate on standard datasets and are compared to the results obtained with widely used SVM classifier.

Graphical Model-Based Vertebra Identification from X-Ray Image(s)

Automated identification of vertebrae from X-ray image(s) is an important step for various medical image computing tasks such as 2D/3D rigid and non-rigid registration. In this chapter we present a graphical model-based solution for automated vertebra identification from X-ray image(s). Our solution does not ask for a training process using training data and has the capability to automatically determine the number of vertebrae visible in the image(s). This is achieved by combining a graphical model-based maximum a posterior probability (MAP) estimate with a mean-shift based clustering. Experiments conducted on simulated X-ray images as well as on a low-dose low quality X-ray spinal image of a scoliotic patient verified its performance.

Non-rigid Free-Form 2D-3D Registration Using Statistical Deformation Model

This paper presents a non-rigid free-from 2D-3D registration approach using statistical deformation model (SDM). In our approach the SDM is first constructed from a set of training data using a non-rigid registration algorithm based on b-spline free-form deformation to encode a priori information about the underlying anatomy. A novel intensity-based non-rigid 2D-3D registration algorithm is then presented to iteratively fit the 3D b-spline-based SDM to the 2D X-ray images of an unseen subject, which requires a computationally expensive inversion of the instantiated deformation in each iteration. In this paper, we propose to solve this challenge with a fast B-spline pseudo-inversion algorithm that is implemented on graphics processing unit (GPU). Experiments conducted on C-arm and X-ray images of cadaveric femurs demonstrate the efficacy of the present approach.

Fluxes and meteorological observations on Spitsbergen

Arctic permafrost may be adversely affected by climate change in a number of ways, so that establishing a world-wide monitoring program seems imperative. This thesis evaluates possibilities for permafrost monitoring at the example of a permafrost site on Svalbard, Norway. An energy balance model for permafrost temperatures is developed that evaluates the different components of the surface energy budget in analogy to climate models. The surface energy budget, consisting of radiation components, sensible and latent heat fluxes as well as the ground heat flux, is measured over the course of one year, which has not been accomplished for arctic land areas so far. A considerable small-scale heterogeneity of the summer surface temperature is observed in long-term measurements with a thermal imaging system, which can be reproduced in the energy balance model. The model can also simulate the impact of different snow depths on the soil temperature, that has been documented in field measurements. Furthermore, time series of terrestrial surface temperature measurements are compared to satellite-borne measurements, for which a significant cold-bias is observed during winter. Finally, different possibilities for a world-wide monitoring scheme are assessed. Energy budget models can incorporate different satellite data sets as training data sets for parameter estimation, so that they may constitute an alternative to purely satellite-based schemes.

Lazy Lasso for local regression

Locally weighted regression is a technique that predicts the response for new data items from their neighbors in the training data set, where closer data items are assigned higher weights in the prediction. However, the original method may suffer from overfitting and fail to select the relevant variables. In this paper we propose combining a regularization approach with locally weighted regression to achieve sparse models. Specifically, the lasso is a shrinkage and selection method for linear regression. We present an algorithm that embeds lasso in an iterative procedure that alternatively computes weights and performs lasso-wise regression. The algorithm is tested on three synthetic scenarios and two real data sets. Results show that the proposed method outperforms linear and local models for several kinds of scenarios

Estimating Perceived Video Quality from Objective Parameters in Video over IP Services

In Video over IP services, perceived video quality heavily depends on parameters such as video coding and network Quality of Service. This paper proposes a model for the estimation of perceived video quality in video streaming and broadcasting services that combines the aforementioned parameters with other that depend mainly on the information contents of the video sequences. These fitting parameters are derived from the Spatial and Temporal Information contents of the sequences. This model does not require reference to the original video sequence so it can be used for online, real-time monitoring of perceived video quality in Video over IP services. Furthermore, this paper proposes a measurement workbench designed to acquire both training data for model fitting and test data for model validation. Preliminary results show good correlation between measured and predicted values.

Automatic Categorization for Improving Spanish into Spanish Sign Language Machine Translation

This paper describes a preprocessing module for improving the performance of a Spanish into Spanish Sign Language (Lengua de Signos Espanola: LSE) translation system when dealing with sparse training data. This preprocessing module replaces Spanish words with associated tags. The list with Spanish words (vocabulary) and associated tags used by this module is computed automatically considering those signs that show the highest probability of being the translation of every Spanish word. This automatic tag extraction has been compared to a manual strategy achieving almost the same improvement. In this analysis, several alternatives for dealing with non-relevant words have been studied. Non-relevant words are Spanish words not assigned to any sign. The preprocessing module has been incorporated into two well-known statistical translation architectures: a phrase-based system and a Statistical Finite State Transducer (SFST). This system has been developed for a specific application domain: the renewal of Identity Documents and Driver's License. In order to evaluate the system a parallel corpus made up of 4080 Spanish sentences and their LSE translation has been used. The evaluation results revealed a significant performance improvement when including this preprocessing module. In the phrase-based system, the proposed module has given rise to an increase in BLEU (Bilingual Evaluation Understudy) from 73.8% to 81.0% and an increase in the human evaluation score from 0.64 to 0.83. In the case of SFST, BLEU increased from 70.6% to 78.4% and the human evaluation score from 0.65 to 0.82.

Neutron spectrometry using artificial neural networks for a bonner sphere spectrometer with 3He detector

Neutron spectra unfolding and dose equivalent calculation are complicated tasks in radiation protection, are highly dependent of the neutron energy, and a precise knowledge on neutron spectrometry is essential for all dosimetry-related studies as well as many nuclear physics experiments. In previous works have been reported neutron spectrometry and dosimetry results, by using the ANN technology as alternative solution, starting from the count rates of a Bonner spheres system with a LiI(Eu) thermal neutrons detector, 7 polyethylene spheres and the UTA4 response matrix with 31 energy bins. In this work, an ANN was designed and optimized by using the RDANN methodology for the Bonner spheres system used at CIEMAT Spain, which is composed of a He neutron detector, 12 moderator spheres and a response matrix for 72 energy bins. For the ANN design process a neutrons spectra catalogue compiled by the IAEA was used. From this compilation, the neutrons spectra were converted from lethargy to energy spectra. Then, the resulting energy ?uence spectra were re-binned by using the MCNP code to the corresponding energy bins of the He response matrix before mentioned. With the response matrix and the re-binned spectra the counts rate of the Bonner spheres system were calculated and the resulting re-binned neutrons spectra and calculated counts rate were used as the ANN training data set.

Modelling seed germination in forest tree species through survival analysis. The Pinus pinea L. case study

The direct application of existing models for seed germination may often be inadequate in the context of ecology and forestry germination experiments. This is because basic model assumptions are violated and variables available to forest managers are rarely used. In this paper, we present a method which addresses the aforementioned shortcomings. The approach is illustrated through a case study of Pinus pinea L. Our findings will also shed light on the role of germination in the general failure of natural regeneration in managed forests of this species. The presented technique consists of a mixed regression model based on survival analysis. Climate and stand covariates were tested. Data for fitting the model were gathered from a 5-year germination experiment in a mature, managed P. pinea stand in the Northern Plateau of Spain in which two different stand densities can be found. The model predictions proved to be unbiased and highly accurate when compared with the training data. Germination in P. pinea was controlled through thermal variables at stand level. At microsite level, low densities negatively affected the probability of germination. A time-lag in the response was also detected. Overall, the proposed technique provides a reliable alternative to germination modelling in ecology/forestry studies by using accessible/ suitable variables. The P. pinea case study highlights the importance of producing unbiased predictions. In this species, the occurrence and timing of germination suggest a very different regeneration strategy from that understood by forest managers until now, which may explain the high failure rate of natural regeneration in managed stands. In addition, these findings provide valuable information for the management of P. pinea under climate-change conditions.

Low-resource language recognition using a fusion of phoneme posteriorgram counts, acoustic and glottal-based i-vectors

This paper presents a description of our system for the Albayzin 2012 LRE competition. One of the main characteristics of this evaluation was the reduced number of available files for training the system, especially for the empty condition where no training data set was provided but only a development set. In addition, the whole database was created from online videos and around one third of the training data was labeled as noisy files. Our primary system was the fusion of three different i-vector based systems: one acoustic system based on MFCCs, a phonotactic system using trigrams of phone-posteriorgram counts, and another acoustic system based on RPLPs that improved robustness against noise. A contrastive system that included new features based on the glottal source was also presented. Official and postevaluation results for all the conditions using the proposed metrics for the evaluation and the Cavg metric are presented in the paper.

Higher-order regularization and morphological techniques for image segmentation

La segmentación de imágenes es un campo importante de la visión computacional y una de las áreas de investigación más activas, con aplicaciones en comprensión de imágenes, detección de objetos, reconocimiento facial, vigilancia de vídeo o procesamiento de imagen médica. La segmentación de imágenes es un problema difícil en general, pero especialmente en entornos científicos y biomédicos, donde las técnicas de adquisición imagen proporcionan imágenes ruidosas. Además, en muchos de estos casos se necesita una precisión casi perfecta. En esta tesis, revisamos y comparamos primero algunas de las técnicas ampliamente usadas para la segmentación de imágenes médicas. Estas técnicas usan clasificadores a nivel de pixel e introducen regularización sobre pares de píxeles que es normalmente insuficiente. Estudiamos las dificultades que presentan para capturar la información de alto nivel sobre los objetos a segmentar. Esta deficiencia da lugar a detecciones erróneas, bordes irregulares, configuraciones con topología errónea y formas inválidas. Para solucionar estos problemas, proponemos un nuevo método de regularización de alto nivel que aprende información topológica y de forma a partir de los datos de entrenamiento de una forma no paramétrica usando potenciales de orden superior. Los potenciales de orden superior se están popularizando en visión por computador, pero la representación exacta de un potencial de orden superior definido sobre muchas variables es computacionalmente inviable. Usamos una representación compacta de los potenciales basada en un conjunto finito de patrones aprendidos de los datos de entrenamiento que, a su vez, depende de las observaciones. Gracias a esta representación, los potenciales de orden superior pueden ser convertidos a potenciales de orden 2 con algunas variables auxiliares añadidas. Experimentos con imágenes reales y sintéticas confirman que nuestro modelo soluciona los errores de aproximaciones más débiles. Incluso con una regularización de alto nivel, una precisión exacta es inalcanzable, y se requeire de edición manual de los resultados de la segmentación automática. La edición manual es tediosa y pesada, y cualquier herramienta de ayuda es muy apreciada. Estas herramientas necesitan ser precisas, pero también lo suficientemente rápidas para ser usadas de forma interactiva. Los contornos activos son una buena solución: son buenos para detecciones precisas de fronteras y, en lugar de buscar una solución global, proporcionan un ajuste fino a resultados que ya existían previamente. Sin embargo, requieren una representación implícita que les permita trabajar con cambios topológicos del contorno, y esto da lugar a ecuaciones en derivadas parciales (EDP) que son costosas de resolver computacionalmente y pueden presentar problemas de estabilidad numérica. Presentamos una aproximación morfológica a la evolución de contornos basada en un nuevo operador morfológico de curvatura que es válido para superficies de cualquier dimensión. Aproximamos la solución numérica de la EDP de la evolución de contorno mediante la aplicación sucesiva de un conjunto de operadores morfológicos aplicados sobre una función de conjuntos de nivel. Estos operadores son muy rápidos, no sufren de problemas de estabilidad numérica y no degradan la función de los conjuntos de nivel, de modo que no hay necesidad de reinicializarlo. Además, su implementación es mucho más sencilla que la de las EDP, ya que no requieren usar sofisticados algoritmos numéricos. Desde un punto de vista teórico, profundizamos en las conexiones entre operadores morfológicos y diferenciales, e introducimos nuevos resultados en este área. Validamos nuestra aproximación proporcionando una implementación morfológica de los contornos geodésicos activos, los contornos activos sin bordes, y los turbopíxeles. En los experimentos realizados, las implementaciones morfológicas convergen a soluciones equivalentes a aquéllas logradas mediante soluciones numéricas tradicionales, pero con ganancias significativas en simplicidad, velocidad y estabilidad. ABSTRACT Image segmentation is an important field in computer vision and one of its most active research areas, with applications in image understanding, object detection, face recognition, video surveillance or medical image processing. Image segmentation is a challenging problem in general, but especially in the biological and medical image fields, where the imaging techniques usually produce cluttered and noisy images and near-perfect accuracy is required in many cases. In this thesis we first review and compare some standard techniques widely used for medical image segmentation. These techniques use pixel-wise classifiers and introduce weak pairwise regularization which is insufficient in many cases. We study their difficulties to capture high-level structural information about the objects to segment. This deficiency leads to many erroneous detections, ragged boundaries, incorrect topological configurations and wrong shapes. To deal with these problems, we propose a new regularization method that learns shape and topological information from training data in a nonparametric way using high-order potentials. High-order potentials are becoming increasingly popular in computer vision. However, the exact representation of a general higher order potential defined over many variables is computationally infeasible. We use a compact representation of the potentials based on a finite set of patterns learned fromtraining data that, in turn, depends on the observations. Thanks to this representation, high-order potentials can be converted into pairwise potentials with some added auxiliary variables and minimized with tree-reweighted message passing (TRW) and belief propagation (BP) techniques. Both synthetic and real experiments confirm that our model fixes the errors of weaker approaches. Even with high-level regularization, perfect accuracy is still unattainable, and human editing of the segmentation results is necessary. The manual edition is tedious and cumbersome, and tools that assist the user are greatly appreciated. These tools need to be precise, but also fast enough to be used in real-time. Active contours are a good solution: they are good for precise boundary detection and, instead of finding a global solution, they provide a fine tuning to previously existing results. However, they require an implicit representation to deal with topological changes of the contour, and this leads to PDEs that are computationally costly to solve and may present numerical stability issues. We present a morphological approach to contour evolution based on a new curvature morphological operator valid for surfaces of any dimension. We approximate the numerical solution of the contour evolution PDE by the successive application of a set of morphological operators defined on a binary level-set. These operators are very fast, do not suffer numerical stability issues, and do not degrade the level set function, so there is no need to reinitialize it. Moreover, their implementation is much easier than their PDE counterpart, since they do not require the use of sophisticated numerical algorithms. From a theoretical point of view, we delve into the connections between differential andmorphological operators, and introduce novel results in this area. We validate the approach providing amorphological implementation of the geodesic active contours, the active contours without borders, and turbopixels. In the experiments conducted, the morphological implementations converge to solutions equivalent to those achieved by traditional numerical solutions, but with significant gains in simplicity, speed, and stability.

Detecting false testimonies in reputation systems using self-organizing maps

It has been demonstrated that rating trust and reputation of individual nodes is an effective approach in distributed environments in order to improve security, support decision-making and promote node collaboration. Nevertheless, these systems are vulnerable to deliberate false or unfair testimonies. In one scenario, the attackers collude to give negative feedback on the victim in order to lower or destroy its reputation. This attack is known as bad mouthing attack. In another scenario, a number of entities agree to give positive feedback on an entity (often with adversarial intentions). This attack is known as ballot stuffing. Both attack types can significantly deteriorate the performances of the network. The existing solutions for coping with these attacks are mainly concentrated on prevention techniques. In this work, we propose a solution that detects and isolates the abovementioned attackers, impeding them in this way to further spread their malicious activity. The approach is based on detecting outliers using clustering, in this case self-organizing maps. An important advantage of this approach is that we have no restrictions on training data, and thus there is no need for any data pre-processing. Testing results demonstrate the capability of the approach in detecting both bad mouthing and ballot stuffing attack in various scenarios.

Towards speaking style transplantation in speech synthesis

One of the biggest challenges in speech synthesis is the production of naturally sounding synthetic voices. This means that the resulting voice must be not only of high enough quality but also that it must be able to capture the natural expressiveness imbued in human speech. This paper focus on solving the expressiveness problem by proposing a set of different techniques that could be used for extrapolating the expressiveness of proven high quality speaking style models into neutral speakers in HMM-based synthesis. As an additional advantage, the proposed techniques are based on adaptation approaches, which means that they can be used with little training data (around 15 minutes of training data are used in each style for this paper). For the final implementation, a set of 4 speaking styles were considered: news broadcasts, live sports commentary, interviews and parliamentary speech. Finally, the implementation of the 5 techniques were tested through a perceptual evaluation that proves that the deviations between neutral and speaking style average models can be learned and used to imbue expressiveness into target neutral speakers as intended.