34 resultados para Self-supervised learning
em BORIS: Bern Open Repository and Information System - Berna - Suiça
Resumo:
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.
Resumo:
In contrast to preoperative brain tumor segmentation, the problem of postoperative brain tumor segmentation has been rarely approached so far. We present a fully-automatic segmentation method using multimodal magnetic resonance image data and patient-specific semi-supervised learning. The idea behind our semi-supervised approach is to effectively fuse information from both pre- and postoperative image data of the same patient to improve segmentation of the postoperative image. We pose image segmentation as a classification problem and solve it by adopting a semi-supervised decision forest. The method is evaluated on a cohort of 10 high-grade glioma patients, with segmentation performance and computation time comparable or superior to a state-of-the-art brain tumor segmentation method. Moreover, our results confirm that the inclusion of preoperative MR images lead to a better performance regarding postoperative brain tumor segmentation.
Resumo:
Facial nerve segmentation plays an important role in surgical planning of cochlear implantation. Clinically available CBCT images are used for surgical planning. However, its relatively low resolution renders the identification of the facial nerve difficult. In this work, we present a supervised learning approach to enhance facial nerve image information from CBCT. A supervised learning approach based on multi-output random forest was employed to learn the mapping between CBCT and micro-CT images. Evaluation was performed qualitatively and quantitatively by using the predicted image as input for a previously published dedicated facial nerve segmentation, and cochlear implantation surgical planning software, OtoPlan. Results show the potential of the proposed approach to improve facial nerve image quality as imaged by CBCT and to leverage its segmentation using OtoPlan.
Resumo:
Patient-specific biomechanical models including local bone mineral density and anisotropy have gained importance for assessing musculoskeletal disorders. However the trabecular bone anisotropy captured by high-resolution imaging is only available at the peripheral skeleton in clinical practice. In this work, we propose a supervised learning approach to predict trabecular bone anisotropy that builds on a novel set of pose invariant feature descriptors. The statistical relationship between trabecular bone anisotropy and feature descriptors were learned from a database of pairs of high resolution QCT and clinical QCT reconstructions. On a set of leave-one-out experiments, we compared the accuracy of the proposed approach to previous ones, and report a mean prediction error of 6% for the tensor norm, 6% for the degree of anisotropy and 19◦ for the principal tensor direction. These findings show the potential of the proposed approach to predict trabecular bone anisotropy from clinically available QCT images.
Resumo:
Finite element (FE) analysis is an important computational tool in biomechanics. However, its adoption into clinical practice has been hampered by its computational complexity and required high technical competences for clinicians. In this paper we propose a supervised learning approach to predict the outcome of the FE analysis. We demonstrate our approach on clinical CT and X-ray femur images for FE predictions ( FEP), with features extracted, respectively, from a statistical shape model and from 2D-based morphometric and density information. Using leave-one-out experiments and sensitivity analysis, comprising a database of 89 clinical cases, our method is capable of predicting the distribution of stress values for a walking loading condition with an average correlation coefficient of 0.984 and 0.976, for CT and X-ray images, respectively. These findings suggest that supervised learning approaches have the potential to leverage the clinical integration of mechanical simulations for the treatment of musculoskeletal conditions.
Resumo:
Background: Residents demonstrate a broad range of performance levels for clinical skills, with some at an inadequate level. Adequate self-assessment is important for life long learning. However, its accuracy is questioned extensively. The aim of this study was to evaluate how far the residents’ self-assessment predicts their performance in an expert assessment of emergency skills. Summary of work: Twelve skills were identified as being relevant for the emergency duties of residents in smaller hospitals. Fifteen first-year residents from the departments of internal medicine and general surgery at a district hospital rated their performance on a questionnaire (self-assessment). This was followed by a structured, practical in vivo assessment by an anaesthesiologist (expert assessment). For both, a visual analogue scale from 0 to 10 was used, on which 0 stands for novice and 10 for expert. Predictive validity was described by Spearman’s correlation, which was significant in 3 out of 12 skills only. Median correlation (r) was 0.50 (range 0.16 to 0.93). Conclusion: At the beginning of postgraduate training, self-assessment alone is not sufficient to guide self-directed learning. Take-home message: At the beginning of their residency, physicians need structured feedback in emergency skills which can be offered by anaesthesiologists.
Resumo:
In der Selbstbestimmungstheorie werden neben der intrinsischen Motivation verschiedene Formen extrinsischer Motivation unterschieden. Dabei wird die Freude am Lernen als positive, aktivierende Emotion, die mit positiven Lernhandlungen und hoher Leistung verknüpft ist, für die intrinsische Motivation als konstitutiv betrachtet. Der Zusammenhang zu den anderen Motivationsformen hingegen ist bisher empirisch weitgehend ungeklärt. An diesem Punkt setzt die vorliegende Studie an. Es wurden 356 Schüler und Schülerinnen aus österreichischen Hauptschulen zu zwei Messzeitpunkten (6. und 7. Schulstufe) mittels Fragebögen zu ihrer Motivation, ihrer Lernfreude, der Mitarbeit und der Leistung in der Schule befragt. Die Ergebnisse aus Pfadanalysen bestätigen die positive Beziehung zwischen der identifizierten Motivation und die negative Beziehung zwischen der externalen Motivation und der Lernfreude. Die introjizierte Regulation korreliert in der 7. Schulstufe schwach positiv, in der 6. Schulstufe nicht mit der Lernfreude. Die Mitarbeit weist sowohl positive Bezüge zur Lernfreude als auch zur identifizierten und introjizierten Motivation, jedoch negative Bezüge zur externalen Motivation auf. Eine hohe Mitarbeit ist förderlich für die Leistung.
Resumo:
Contemporary models of self-regulated learning emphasize the role of distal motivational factors for student's achievement, on the one side, and the proximal role of metacognitive monitoring and control for learning and test outcomes, on the other side. In the present study, two larger samples of elementary school children (9- and 11-year-olds) were included and their mastery-oriented motivation, metacognitive monitoring and control skills were integrated into structural equation models testing and comparing the relative impact of these different constituents for self-regulated learning. For one, results indicate that the factorial structure of monitoring, control and mastery motivation was invariant across the two age groups. Of specific interest was the finding that there were age-dependent structural links between monitoring, control, and test performance (closer links in the older compared to the younger children), with high confidence yielding a direct and positive effect on test performance and a direct and negative effect on adequate control behavior in the achievement test. Mastery-oriented motivation was not found to be substantially associated with monitoring (confidence), control (detection and correction of errors), or test performance underlining the importance of proximal, metacognitive factors for test performance in elementary school children.
Children's deliberate memory development: The contribution of strategies and metacognitive processes
Resumo:
This chapter focus is laid on the development of memory skills when children are confronted with a task or a situation in which learning or remembering certain target information is crucial. It presents important milestones toward self-regulated learning skills. The chapter discusses precursors of later strategic behaviors and metacognitive skills, the distinct research methods suitable to assessing early indicators of deliberate memory skills, as well as their importance for the emerging memory skills. It outlines the challenges arising from the application of deliberate memory skills in naturalistic, complex task contexts. The chapter adopted an explicit developmental perspective of memory strategies and metacognition in deliberate memory situations.
Resumo:
Location-awareness indoors will be an inseparable feature of mobile services/applications in future wireless networks. Its current ubiquitous availability is still obstructed by technological challenges and privacy issues. We propose an innovative approach towards the concept of indoor positioning with main goal to develop a system that is self-learning and able to adapt to various radio propagation environments. The approach combines estimation of propagation conditions, subsequent appropriate channel modelling and optimisation feedback to the used positioning algorithm. Main advantages of the proposal are decreased system set-up effort, automatic re-calibration and increased precision.
Resumo:
OBJECTIVES Evidence increases that cognitive failure may be used to screen for drivers at risk. Until now, most studies have relied on driving learners. This exploratory pilot study examines self-report of cognitive failure in driving beginners and error during real driving as observed by driving instructors. METHODS Forty-two driving learners of 14 driving instructors filled out a work-related cognitive failure questionnaire. Driving instructors observed driving errors during the next driving lesson. In multiple linear regression analysis, driving errors were regressed on cognitive failure with the number of driving lessons as an estimator of driving experience controlled. RESULTS Higher cognitive failure predicted more driving errors (p < .01) when age, gender and driving experience were controlled in analysis. CONCLUSIONS Cognitive failure was significantly associated with observed driving errors. Systematic research on cognitive failure in driving beginners is recommended.