Las dificultades del aprendizaje se cuentan entre las principales razones de la deserción escolar, siendo éste un problema a nivel global. El diagnóstico temprano de estas dificultades es primordial para abordar un tratamiento de manera temprana y un acompañamiento adecuado. Hoy en día el diagnóstico de estas dificultades se lleva a cabo por psicopedagogos y/o psicólogos evaluando la resolución de una batería de tests. La relación 1 a 1 entre una persona a diagnosticar y el profesional dificulta la tarea de estudios masivos para detectar tempranamente a niños en edad escolar que sufran alguna dificultad específica del aprendizaje. Un síntoma estudiado para detectar procesos cognitivos anómalos son comportamientos atípicos en los movimientos oculares. Con la tecnología de acceso masivo se pueden registrar los movimientos oculares, de manera sencilla y económica, con dispositivos de seguimiento ocular llamados por su nombre en inglés eye trackers. En esta tesis se desarrollaron herramientas para el estudio de las señales de los dispositivos mencionados. Una de estas herramientas es clave para estudiar los movimientos oculares y es la clasificación de estos en dos grandes grupos llama- dos fijaciones y sacadas. Las características estadísticas que poseen estos movimientos están íntimamente relacionados con procesos cognitivos. Otra herramienta desarrolla- da es el cálculo de diferentes cantidades de carácter estadístico que permiten describir en pocos números múltiples propiedades de señales obtenidas con los eye trackers. Una dificultad específica del aprendizaje que padece alrededor del 10 % de la población es la dislexia. Ésta es un trastorno con el que, aún hoy, hay controversia al momento de definirlo, pero que es una dificultad que afecta el desarrollo el desarrollo de la alfabetización y las habilidades relacionadas con el lenguaje. En esta tesis se estudiaron los movimientos oculares de niños que fueron diagnosticados con dislexia y de niños de un grupo control que han tenido un desarrollo típico. Describimos la dinámica de la trayectoria de la mirada de los niños al leer con un modelo estocástico basado en lo que se conoce como Continuous Time Random Walk (CTRW). Con dicho modelo fuimos capaces de generar señales sintéticas con iguales características que las señales reales. La generación de señales sintéticas es una técnica clave en el aprendizaje automático, o aprendizaje de máquina (machine learning) y es importante porque ayuda a mejorar el rendimiento de los modelos, combatir el sobre-ajuste y manejar la escasez de datos. Con datos reales hemos probado una serie de clasificadores de disléxicos y no disléxicos utilizando herramientas de machine learning. Dos de estos métodos tuvieron una alta precisión. Tener herramientas para analizar movimientos oculares podría ayudar en el diagnóstico rápido de dificultades del aprendizaje. A su vez podría ayudar a evaluar de forma masiva a niños en sus colegios y así bajar de forma drástica la deserción escolar. / Learning difficulties are among the main reasons for school dropout, this being a global problem. The early diagnosis of these difficulties is essential to address early treatment and adequate follow-up. Today the diagnosis of these difficulties is carried out by educational psychologists and/or psychologists evaluating the resolution of a battery of tests. The 1 to 1 relationship between a person to be diagnosed and the professional makes the task of massive studies difficult to detect early school-age children who suffer from a specific learning difficulty. A symptom studied to detect abnormal cognitive processes are atypical behaviors in eye movements. With mass access technology, eye movements can be recorded easily and cheaply with eye tracking devices. In this thesis, tools were developed to study the signals of the mentioned devices. One of these tools is key to studying eye movements and it is the classification of these into two large groups called fixations and saccades. The statistical characteristics that these movements have are closely related to cognitive processes. Another tool developed is the calculation of different quantities of a statistical nature that allow multiple properties of signals obtained with eye trackers to be described in a few numbers. A specific learning difficulty that about 10 % of the population suffers from is dyslexia. This is a disorder with which, even today, there is controversy when defining it, but it is a difficulty that affects the development of literacy and language-related skills. In this thesis, the eye movements of children who were diagnosed with dyslexia and of children from a control group who have had a typical development were studied. We describe the dynamics of children’s gaze trajectory when reading with a stochastic model based on what is known as Continuous Time Random Walk (CTRW). With this model we were able to generate synthetic signals with the same characteristics as the real signals. The generation of synthetic signals is a key technique in machine learning, or machine learning, and is important because it helps improve model performance, combat overfitting, and handle data sparseness. With real data we have tested a series of dyslexic and non-dyslexic classifiers using machine learning tools. Two of these methods had high precision. Having tools to analyze eye movements could help in the rapid diagnosis of learning difficulties. In turn, it could help to massively evaluate children in their schools and thus drastically reduce school dropouts, with all the benefits that this entails for society.

No description available.

No description available.

No description available.

This thesis presents a generalized Bayesian framework for a mobile robot to localize itself and a target, while building a map of the environment. The proposed technique builds upon the Bayesian Simultaneous Robot Localization and Mapping (SLAM) method, to allow the robot to localize itself and the environment using map features or landmarks in close proximity. The target feature is distinguished from the rest of features since the robot has to navigate to its location and thus needs to be observed from a long distance. The contribution of the proposed approach is on enabling the robot to track a target object or region, using a multi-stage technique. In the first stage, the target state is corrected sequentially to the robot correction in the Recursive Bayesian Estimation. In the second stage, with the target being closer, the target state is corrected simultaneously with the robot and the landmarks. The process allows the robot's state uncertainty to be propagated into the estimated target's state, bridging the gap between tracking only methods where the target is estimated assuming known observer state and SLAM methods where only landmarks are considered. When the robot is located far, the sequential stage is efficient in tracking the target position while maintaining an accurate robot state using close only features. Also, target belief is always maintained in comparison to temporary tracking methods such as image-tracking. When the robot is closer to the target and most of its field of view is covered by the target, it is shown that simultaneous correction needs to be used in order to minimize robot, target and map entropies in the absence of other landmarks. / M.S. / This thesis presents a generalized framework with the goal of allowing a robot to localize itself and a static target, while building a map of the environment. This map is used as in the Simultaneous Localization and Mapping (SLAM) framework to enhance robot accuracy and with close features. Target, here, is distinguished from the rest of features since the robot has to navigate to its location and thus needs to be continuously observed from a long distance. The contribution of the proposed approach is on enabling the robot to track a target object or region, using a multi-stage technique. In the first stage, the robot and close landmarks are estimated simultaneously and they are both corrected. Using the robot's uncertainty in its estimate, the target state is then estimated sequentially, considering known robot state. That decouples the target estimation from the rest of the process. In the second stage, with the target being closer, target, robot and landmarks are estimated simultaneously. When the robot is located far, the sequential stage is efficient in tracking the target position while maintaining an accurate robot state using close only features. When the robot is closer to the target and most of its field of view is covered by the target, it is shown that simultaneous correction needs to be used in order to minimize robot, target and map uncertainties in the absence of other landmarks.

Studies examining impaired emotion recognition and perceptual processing in autism spectrum disorders (ASD) show inconsistent results (Harms, Martin, & Wallace, 2010; Jemel, Mottron, & Dawson, 2006), and many of these studies include eye tracking data. The current study utilizes a novel task, emotion recognition of a dynamic talking face with sound, to compare children with ASD (n=8; aged 6-10, 7 male) with mental age (MA) and gender matched controls (n=8; aged 4-10, 7 male) on an emotion identification and eye tracking task. Children were asked to watch several short video clips (2.5-5 seconds) portraying the emotions of happy, sad, excited, scared, and angry and identify the emotion portrayed in the video. A mixed factorial ANOVA analysis was conducted to examine group differences in attention when viewing the stimuli. Differences in emotion identification ability were examined using a t-test and Fisher's exact tests of independence. Findings indicated that children with ASD spent less time looking at faces and the mouth region than controls. Additionally, the amount of time children with ASD spent looking at the mouth region predicted better performance on the emotion identification task. The study was underpowered; however, so these results were preliminary and require replication. Results are discussed in relation to natural processing of emotion and social stimuli. <i>[revised ETD per Dean DePauw 10/25/12 GMc]</i> / Master of Science

International Telemetering Conference Proceedings / October 25-28, 1993 / Riviera Hotel and Convention Center, Las Vegas, Nevada / The first telemetry tracking system was desired in 1959 for the space program. Cost was of little concern. The tracking technique used was 3 channel monopulse which is still today, after all these years, the optimum in performance for any type of tracking requirement. Telemetry tracking really got off the ground in the early 1970's with the move from P-Band to S-Band for telemetry. In the design of early tracking systems, performance was on the top of the list, and cost was on the bottom of the list in establishing the design criteria. By the beginning of the 1980's cost was approaching performance in importance. Today, with the demise of the cold war and a considerable reduction in global threats coupled with the state of the world economy, cost has now reached the top of the list. The cost of a telemetry tracking system can be reduced by more than a factor of two by going to a single axis tracking technique. The lowest cost single axis approach heretofore has been the use of a cosecant squared (CSC²) distribution. To improve the efficiency of a single axis system and increase the overhead coverage capability, the use of a dual beam antenna has been widely used as another type of single axis approach. The dual beam technique involves additional costs since two tracking antennas are required. Except for satellite tracking, almost all telemetry tracking is performed at low elevation angles and, like it or not, multipath is there. The multipath fade varies from a few dB, to over 20 dB depending upon the reflecting terrain. Most general purpose systems should be designed for at least a 10 dB multipath fade. For all telemetry tracking applications, the multipath effect is completely negligible at elevation angles greater than 10 degrees. The Augustin-Sullivan Distribution, in effect, fades away the multipath margin as the multipath effect decreases. Because of the multipath phenomenon, an antenna beam should not be shaped at the one dB point as is the case with a CSC² distribution, but only needs to be shaped from somewhere between the 15 - 20 dB level based on the mission requirements. This involves a gain reduction from a pencil beam on the order of 1/2 dB or less, rather than the 3 dB reduction associated with the CSC² distribution. The Augustin-Sullivan distribution does not start shaping the beam until shaping is retired, and shapes the beam for constant altitude coverage from the horizon to zenith. For the first time, coverage is provided from the peak of the beam to directly overhead with a single antenna and a single axis rotator. When GPS information is available from the tracked vehicle, the Augustin-Sullivan distribution, with a single axis rotator and using the GPS-TRAK technique, results in the lowest possible cost alternate to autotracking.

Multiple object tracking is the process of assigning unique and consistent identities to objects throughout a video sequence. A popular approach to multiple object tracking, and object tracking in general, is to use a method called tracking-by-detection. Tracking-by-detection is a two-stage procedure: an object detection algorithm first detects objects in a frame, these objects are then associated with already tracked objects by a tracking algorithm. One of the main concerns of this thesis is to investigate how different object detection algorithms perform on surveillance video supplied by National Forensic Centre. The thesis then goes on to explore how the stand-alone alone performance of the object detection algorithm correlates with overall performance of a tracking-by-detection system. Finally, the thesis investigates how the use of visual descriptors in the tracking stage of a tracking-by-detection system effects performance.  Results presented in this thesis suggest that the capacity of the object detection algorithm is highly indicative of the overall performance of the tracking-by-detection system. Further, this thesis also shows how the use of visual descriptors in the tracking stage can reduce the number of identity switches and thereby increase performance of the whole system.

Target tracking is a broad subject area extensively studied in many engineering disciplines. In this thesis, target tracking implies the temporal estimation of target features such as the target's direction-of-arrival (DOA), the target's boundary pixels in a sequence of images, and/or the target's position in space. For multiple target tracking, we have introduced a new motion model that incorporates an acceleration component along the heading direction of the target. We have also shown that the target motion parameters can be considered part of a more general feature set for target tracking, e.g., target frequencies, which may be unrelated to the target motion, can be used to improve the tracking performance. We have introduced an acoustic multiple-target tracker using a flexible observation model based on an image tracking approach by assuming that the DOA observations might be spurious and that some of the DOAs might be missing in the observation set. We have also addressed the acoustic calibration problem from sources of opportunity such as beacons or a moving source. We have derived and compared several calibration methods for the case where the node can hear a moving source whose position can be reported back to the node. The particle filter, as a recursive algorithm, requires an initialization phase prior to tracking a state vector. The Metropolis-Hastings (MH) algorithm has been used for sampling from intractable multivariate target distributions and is well suited for the initialization problem. Since the particle filter only needs samples around the mode, we have modified the MH algorithm to generate samples distributed around the modes of the target posterior. By simulations, we show that this mode hungry algorithm converges an order of magnitude faster than the original MH scheme. Finally, we have developed a general framework for the joint state-space tracking problem. A proposal strategy for joint state-space tracking using the particle filters is defined by carefully placing the random support of the joint filter in the region where the final posterior is likely to lie. Computer simulations demonstrate improved performance and robustness of the joint state-space when using the new particle proposal strategy.

Signaling Overhead has always been a concern for network operators. LTE offers many improvements aimed at improved network performance and management. This thesis exploit Tracking Area List (TAL) concept in LTE networks. An algorithm to design TAL using UE traces is developed. The performance of TAL design is compared to conventional TA design. Performance is also compared with rule of thumb TAL design which is another approach to designing TAL