检索结果-内蒙古大学图书馆

Metric-DST: Mitigating Selection Bias Through Diversity-Guided Semi-Supervised Metric Learning

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Tepeli, Yasin I. de Wolf, Mathijs Gonçalves, Joana P. Pattern Recognition & Bioinformatics Department of Intelligent Systems EEMCS Faculty Delft University of Technology Delft2628 XE Netherlands

Selection bias poses a critical challenge for fairness in machine learning, as models trained on data that is less representative of the population might exhibit undesirable behavior for underrepresented profiles. Semi-supervised learning strategies like self-training can mitigate selection bias by incorporating unlabeled data into model training to gain further insight into the distribution of the population. However, conventional self-training seeks to include high-confidence data samples, which may reinforce existing model bias and compromise effectiveness. We propose Metric-DST, a diversity-guided self-training strategy that leverages metric learning and its implicit embedding space to counter confidence-based bias through the inclusion of more diverse samples. Metric-DST learned more robust models in the presence of selection bias for generated and real-world datasets with induced bias, as well as a molecular biology prediction task with intrinsic bias. The Metric-DST learning strategy offers a flexible and widely applicable solution to mitigate selection bias and enhance fairness of machine learning models. Copyright © 2024, The Authors. All rights reserved.

关键词： Self-supervised learning

Irregular measurement times in estimating time-varying treatment effects: Categorizing biases and comparing adjustment methods

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Kant, Wouter M.R. Krijthe, Jesse H. Institute for Computing and Information Sciences Radboud University Nijmegen Netherlands Pattern recognition & Bioinformatics Group EEMCS Delft University of Technology Delft Netherlands

To estimate the causal effect of treatments that vary over time from observational data, one must adjust for time-varying confounding. A common procedure to address confounding is the use of inverse probability of treatment weighting methods. However, the timing of covariate measurements is often irregular, which may introduce additional confounding bias as well as selection bias into the causal effect estimate. Two reweighting methods have been proposed to adjust for these biases: time-as-confounder and reweighting by measurement time. However, it is currently not well understood in which situations these irregularly timed measurements induce bias, and how the available reweighting methods compare to each other in different situations. In this work, we provide a complete inventarization of all possible backdoor paths through which bias is induced. Based on these paths, we distinguish three categories of confounding bias by measurement time: direct confounding (DC), confounding through measured variables (CMV), and confounding through unmeasured variables (CUV). These categories differ in the assumptions and reweighting methods necessary to adjust for bias and may occur simultaneously with selection bias. Through simulation studies, we illustrate: 1. Reweighting by measurement time may be used to adjust for selection bias and confounding through measured variables;2. Time-as-confounder may be used to adjust for all categories of confounding bias, but not selection bias;3. In some cases, the use of a combination of both techniques may be used to adjust for both confounding and selection bias. We finally apply the categorization and reweighting methods on the pre-DIVA1 data set. Adjusting for measurement times is crucial in order to avoid bias, and the categorization of biases and techniques that we introduce may help researchers to choose the appropriate analysis. Copyright © 2025, The Authors. All rights reserved.

关键词： Inverse problems

Technologies Supporting Self-Reflection on Social Interactions: A Systematic Review 25

学校读者我要写书评

暂无评论

Technologies Supporting Self-Reflection on Social Interactio...

30th International Conference on Intelligent User Interfaces, IUI 2025

作者： Hao, Chenxu Matej Hrkalovic, Tiffany Balliet, Daniel Hung, Hayley Dudzik, Bernd Delft University of Technology Delft Netherlands Jheronimus Academy of Data Science 's-Hertogenbosch Netherlands Tilburg University Tilburg Netherlands Vrije Universiteit Amsterdam Amsterdam Netherlands Intelligent Systems Delft University of Technology Delft Netherlands Intelligent Systems/Pattern Recognition and Bioinformatics Delft University of Technology Delft Netherlands

ISBN: (纸本)9798400713064

As intelligent technology and applications have become an integral part of nearly all aspects of people's daily lives, many intelligent systems have been designed to help people navigate the complex space of social interactions. One prominent strategy for such intelligent support is providing meaningful Ad Hoc Interventions (ADI), e.g., through timely notifications. An alternative is Technology-Supported Reflection (TSR), e.g., by offering information about activities in one's past for personal insights. In contrast to straight-up interventions, the aim of the latter strategy is not to directly augment human skills but instead support learning and personal growth over time. However, while TSR has seen widespread interest in applications in some areas, such as physical fitness and mental health, its use for improving human social interactions has not yet been systematically explored. Concretely, it is currently unclear 1) what forms of self-reflection systems intend to support, 2) how their different technological components (e.g., data collection, information integration) are involved in providing support, and 3) what common limitations and design challenges they face. In this article, we present the results of a systematic literature review focusing on these questions to provide a structured foundation for targeted research. Concretely, we identified and analysed a collection of 23 relevant papers, each describing a system deploying TSR to support humans with elements of social *** constructed a framework with a set of features to comprehensively describe and analyze the systems that support self-reflection, including their application domains, how they fit into the existing design framework, how they facilitate learning through reflection, how adaptive they are to individual users, and how they were evaluated. Finally, we propose a direction for designing systems that support individual's social interactions through self-reflection in an adaptive manne

关键词： Human computer interaction

Building a Deep Learning Model to Detect Osteoporosis from Dental Panoramic X-Ray Image

学校读者我要写书评

暂无评论

Open Journal of Applied Sciences 2024年第12期14卷 3480-3489页

作者： Farah Hassan Brangakgi Yasser Khadra Department of Physiology Faculty of Medicine Damascus University Damascus Syria Bioinformatics Syrian Virtual University Damascus Syria Medical Digital Image Processing and Pattern Recognition University of Lyon Lyon France

The project discusses the development of a deep learning model to detect osteoporosis from dental panoramic X-Ray images. It provides an in-depth understanding of human bone structure, osteoporosis, its symptoms, causes, prevalence, and risk factors. The project also explains bone density measurement using dual-energy X-ray absorptiometry (DEXA) and the application of artificial intelligence (AI) and machine learning (ML) in medical imaging. The study uses panoramic dental X-rays to evaluate AI technology in dental imaging and classification of mandible inferior cortical based on Klemetti and Kolmakow criteria. The model architecture consists of convolutional, pooling, fully connected, ReLU, and Softmax layers. Dropout and early stopping are added to the model. The training process uses the train-test approach with 100 epochs and a batch size of 32, and performance evaluation measures such as accuracy, sensitivity, specificity, and F1-score are used to assess the classifier’s performance. The findings and methodology provide a comprehensive understanding of the application of deep learning in the detection of osteoporosis from dental panoramic X-Ray images, and the study demonstrates a robust approach to implementing AI in medical imaging for osteoporosis detection.

关键词： Deep Learning CNN Convolutional Neural Network Python Osteoporosis Dental Panoramic X-Ray

Self-Supervised PPG Representation Learning Shows High Inter-Subject Variability

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Ghorbani, Ramin Reinders, Marcel T.J. Tax, David M.J. Pattern Recognition and Bioinformatics group Delft University of Technology Delft Netherlands

With the progress of sensor technology in wearables, the collection and analysis of PPG signals are gaining more interest. Using Machine Learning, the cardiac rhythm corresponding to PPG signals can be used to predict different tasks such as activity recognition, sleep stage detection, or more general health status. However, supervised learning is often limited by the amount of available labeled data, which is typically expensive to obtain. To address this problem, we propose a Self-Supervised Learning (SSL) method with a pretext task of signal reconstruction to learn an informative generalized PPG representation. The performance of the proposed SSL framework is compared with two fully supervised baselines. The results show that in a very limited label data setting (10 samples per class or less), using SSL is beneficial, and a simple classifier trained on SSL-learned representations outperforms fully supervised deep neural networks. However, the results reveal that the SSL-learned representations are too focused on encoding the subjects. Unfortunately, there is high inter-subject variability in the SSL-learned representations, which makes working with this data more challenging when labeled data is scarce. The high inter-subject variability suggests that there is still room for improvements in learning representations. In general, the results suggest that SSL may pave the way for the broader use of machine learning models on PPG data in label-scarce regimes. © 2022, CC BY.

关键词： pattern recognition

Improving performance of heart rate time series classification by grouping subjects

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Beekhuizen, Michael Naseri, Arman Tax, David van der Bilt, Ivo Reinders, Marcel Pattern Recognition and Bioinformatics Delft University of Technology Delft Netherlands Department of Cardiology Haga Teaching Hospital The Hague Netherlands

Unlike the more commonly analyzed ECG or PPG data for activity classification, heart rate time series data is less detailed, often noisier and can contain missing data points. Using the BigIdeasLab_STEP dataset, which includes heart rate time series annotated with specific tasks performed by individuals, we sought to determine if general classification was achievable. Our analyses showed that the accuracy is sensitive to the choice of window/stride size. Moreover, we found variable classification performances between subjects due to differences in the physical structure of their hearts. Various techniques were used to minimize this variability. First of all, normalization proved to be a crucial step and significantly improved the performance. Secondly, grouping subjects and performing classification inside a group helped to improve performance and decrease inter-subject variability. Finally, we show that including handcrafted features as input to a deep learning (DL) network improves the classification performance further. Together, these findings indicate that heart rate time series can be utilized for classification tasks like predicting activity. However, normalization or grouping techniques need to be chosen carefully to minimize the issue of subject variability. © 2023, CC BY.

关键词： Time series

Effects of Social Behaviours in Online Video Games on Team Trust 22

学校读者我要写书评

暂无评论

Effects of Social Behaviours in Online Video Games on Team T...

9th ACM SIGCHI Annual Symposium on Computer-Human Interaction in Play, CHI PLAY 2022

作者： Van Rhenen, Jan-Willem Centeio Jorge, Carolina Matej Hrkalovic, Tiffany Dudzik, Bernd Intelligent Systems Delft University Of Technology Netherlands Interactive Intelligence Delft University Of Technology Netherlands Vrije Universiteit Amsterdam Amsterdam North Holland Delft University Of Technology South Holland Delft Netherlands Pattern Recognition And Bioinformatics Delft University Of Technology Netherlands

ISBN: (纸本)9781450392112

In competitive multiplayer online video games, teamwork is of utmost importance, implying high levels of interdependence between the joint outcomes of players. When engaging in such interdependent interactions, humans rely on trust to facilitate coordination of their individual behaviours. However, online games often take place between teams of strangers, with individual members having little to no information about each other than what they observe throughout the interaction itself. A better understanding of the social behaviours that are used by players to form trust could not only facilitate richer gaming experiences, but could also lead to insights about team interactions. As such, this paper presents a first step towards understanding how and which types of in-game behaviour relate to trust formation. In particular, we investigate a) which in-game behaviour were relevant for trust formation (first part of the study) and b) how they relate to the reported player's trust in their teammates (the second part of the study). The first part consisted of interviews with League of Legends players in order to create a taxonomy of in-game behaviours relevant for trust formation. As for the second part, we ran a small-scale pilot study where participants played the game and then answered a questionnaire to measure their trust in their teammates. Our preliminary results present a taxonomy of in-game behaviours which can be used to annotate the games regarding trust behaviours. Based on the pilot study, the list of behaviours could be extended as to improve the results. These findings can be used to research the role of trust formation in teamwork. © 2022 Owner/Author.

关键词： Taxonomies

Heart disease detection using an acceleration-deceleration curve-based neural network with consumer-grade smartwatch data

学校读者我要写书评

暂无评论

Heliyon 2024年第21期10卷 e39927页

作者： Naseri, Arman Tax, David M.J. Reinders, Marcel van der Bilt, Ivo Department of Cardiology Haga Teaching Hospital The Hague Netherlands Pattern Recognition and Bioinformatics Delft University of Technology Delft Netherlands Department of Cardiology University Medical Center Utrecht Utrecht Netherlands

Cardiovascular disease (CVD) is the most important cause of morbidity and mortality worldwide. Early detection, prevention or even prediction is of pivotal importance to reduce the burden of cardiovascular disease and its associated costs. Low cost, consumer-grade smartwatches have the potential to revolutionize cardiovascular medicine by enabling continuous monitoring of heart rate and activity. When combined with machine learning(ML), the resulting large amounts of time series data hold the potential of detection, or exclusion of CVD. However, analyzing such large datasets is challenging due to the sparse presence of informative segments. Efficient selection of these segments is essential for developing predictive models for clinical deployment. The objective of this paper was to investigate the potential of an acceleration-deceleration curvebased ML model as a novel clinical indicator for the detection of cardiovascular diseases. We used data from the ME-TIME study;42 participants from which 21 have a cardiovascular disease and 21 are health controls. Data from each subject was normalized to decrease inter-subject variability. A neural network model aggregated predictions per week. We showed that per-subject normalization by the peak value of curves during inactivity, aggregation of model predictions over a week, and using a contrastive loss, resulted in a predictive model with 99 % ± 3 % specificity and 40 % ± 49 % sensitivity on the development set, and 100 % specificity with 67 % ± 47 % sensitivity on the test set. Acceleration-deceleration curves are effective patterns for ruling out the presence of cardiovascular disease, but caution must be taken to properly pre-process the curves and carefully choosing a model that reduces the variability in the extracted curves. © 2024 The Authors

关键词： Atrial fibrillation Ehealth Heart failure Machine learning Mhealth Smartwatch Wearables

POS1052 PREDICTING FIBROMYALGIA OUTCOME FROM A RHEUMATIC-SPECIFIC SURVEY

学校读者我要写书评

暂无评论

Annals of the Rheumatic Diseases 2024年 83卷 573-574页

作者： D. Selani L. Qin F. Zegers L. Klareskog E. Steyerberg S. Le Cessie B. Axnäs M.J.T. Reinders E.B. Van den Akker R. Knevel Delft University of Technology Delft Bioinformatics Lab Pattern Recognition and Bioinformatics Delft Netherlands Leiden University Medical Center Department of Rheumatology Leiden Netherlands Leiden University Medical Center Department of Biomedical Data Sciences Leiden Netherlands Karolinska Institute Stockholm Sweden Leiden University Medical Center Department of Clinical Epidemiology Leiden Netherlands Elsa Science AB Stockholm Sweden Delft University of Technology Delft Bioinformatics Lab Delft Netherlands

Background: Around 3% of the global population suffers from Fibromyalgia (FM) and the average diagnostic delay is six years [1]. It is a hard disease to diagnose as it has symptoms that are common in non-Inflammatory and inflammatory diseases [2]. An outcome classification based on self-reported symptoms could benefit patients, by allowing patients to be referred to specialists sooner. “Rheumatic?” is an online symptom checker originally designed to differentiate patients with immune-mediated from non-immune-mediated rheumatic disease [3], yet it might also aid in the early identification of patients at risk of FM. A previous machine-learning approach trained on “Rheumatic?” data performed moderately well at diagnosis prediction task, with an AUROCS of 0.54 – 0.79 [3]. Objectives: We investigate whether self-reported symptoms collected from “ Rheumatic?” can be used to identify FM diagnosis. Methods: “Rheumatic?” includes a baseline questionnaire about current symptoms and past diagnoses, along with three follow-up surveys (at 3,6 and 12 months) about healthcare visits, medication, and further diagnoses. We selected participants that entered the “Rheumatic?” between July 2021 and July 2022. Each participant’s baseline data was projected into an embedding (Figure 1A) using a pre-trained auto-encoder model. The embedding was then used, along with FM outcomes as labels, to train an XG-Boost classifier (Figure 1B). The data was split 80:20 into a train set and test set. The train set was used to train the auto-encoder model in an unsupervised manner, and the classifier in a supervised manner. The test set was used to get the final model performance metrics. We used the Area Under the Receiver Operating Curve Score (AUROCS) to assess model performance. A separate pipeline was created for every follow-up timepoint. Results: The number of participants at each follow-up changed as not every participant replied to a follow-up survey (Table 1). As expected, participants who r

关键词：