263134 research outputs found
Sort by
Robuuste methoden voor het opsporen van anomalieën met toepassingen voor empirisch onderbouwd beleid
No full textThe project - in cooperation with the Joint Research Centre (JRC) of the European Commission - develops and applies robust statistical techniques for outlier/anomaly detection. The main methodological contributions relate to robust co-clustering and robust regression techniques, addressing both casewise and cellwise outliers. Such methods will are applied to high-dimensional and heterogeneous data provided by the JRC coming from different fields of application, including fraud detection and international trade monitoring.
In more detail, the work consists of four contributions spanning clustering and regression.
First, two robust co-clustering methods are introduced under the Latent Block Model. The RoCo-TRICC approach handles row-wise and column-wise contamination via impartial trimming, while Cell-TRICC is the first method to address cellwise outliers in this framework and incorporates missing data, model selection, and new diagnostic tools. Both methods are validated through simulations and real applications.
The second part of the PhD thesis concerns robust regression. A fraud-detection framework combining MM-regression with spatial autocorrelation measures is proposed for the EU tobacco track-and-trace system. Finally, a robust and regularised Generalised Linear Model for high-dimensional mixed-type data is developed, featuring protection against leverage and vertical outliers, tailored penalties, robust initialisation, and model selection.status: Publishe
Lexically-oriented word alignment for Ancient Greek: a learning-to-rank approach
No full textstatus: Accepte
Accuracy of polarized laser scattering detection of surface/subsurface damage in ground quartz glass
No full textQuartz glass is widely used in high precision optical systems and instruments due to its excellent physical, chemical, and mechanical properties. Developing technology has increasing requirements on the surface integrity of a quartz glass part. However, there is a lack of high-efficiency and precision detection to the surface and subsurface damage in transparent quartz glass, which hinders the improvement of the surface integrity during a machining process. This study proposes a nondestructive dual sensor polarized laser scattering (PLS) system to detect the surface and subsurface damage in quartz glass. One sensor coupled with an integrating sphere is adopted to detect the surface damage. The other sensor is used to detect the subsurface damage. Both the surface and subsurface damage was measured by conventional methods to provide comparisons. Results show that the proposed dual-sensor PLS system represents high accuracies in detecting surface and subsurface damage in quartz glass. The detection errors of surface damage are less than 2%. The detection errors of subsurface damage are about 6%. The PLS signal is sensitive to the damage depth in quartz glass. It represents a constant sensitivity to the damage ranging from sub-micrometer to a few micrometers, which is beneficial to quantitively characterizing and evaluating the damage. Therefore, the dual sensor PLS system could achieve a detection with resolution to submicrometer scale. This study paves the way to the nondestructive detection of the damage in such transparent materials.sponsorship: The study is supported by the National Natural Science Foundation of China (No. 52105453) . (National Natural Science Foundation of China|52105453)status: Published onlin
Optimal s-boxes against alternative operations and linear propagation
No full textsponsorship: M. Calderini and R. Civino are members of INdAM-GNSAGA (Italy) and thankfully acknowledge support by MUR-Italy via PRIN 2022RFAZCJ 'Algebraic methods in cryptanalysis'. R. Civino is supported by the Centre of EXcellence on Connected, Geo-Localized and Cybersecure Vehicles (EX-Emerge) , funded by Italian Government under CIPE resolution n. 70/2017 (Aug. 7, 2017) . R. Invernizzi is supported by the European Research Council (ERC) under the European Union's Horizon 2020 research and innovation programme (grant agreement ISOCRYPT-No. 101020788) and by CyberSecurity Research Flanders with reference number VR20192203. (MUR-Italy|PRIN 2022RFAZCJ, Centre of EXcellence on Connected, Geo-Localized and Cybersecure Vehicles (EX-Emerge) - Italian Government under CIPE resolution|70/2017, European Research Council (ERC)|101020788, CyberSecurity Research Flanders|VR20192203, European Research Council (ERC)|101020788)status: Accepte
Boekbespreking van ‘Tot exemple van anderen’. Heksenvervolging in Zeeland.
No full textstatus: Accepte
Phylogenetic insights into the transmission dynamics of arthropod-borne viruses
No full textArthropod-borne viruses (arboviruses) impose substantial global health and economic burdens, affecting both human and animal populations. These viruses - including dengue, chikungunya, Rift Valley fever, Crimean-Congo haemorrhagic fever and bluetongue viruses - have complex transmission cycles involving vertebrate hosts and arthropod vectors. Their circulation in livestock and wildlife complicate surveillance, as traditional epidemiological approaches rely mainly on human clinical data. Climate change and increasing global interconnectedness are accelerating their emergence and invasion, necessitating a deeper understanding of their ecological and epidemiological dynamics. Advances in genomic surveillance and phylogenetics can provide insights into spatial and temporal patterns of virus transmission that are difficult to obtain through traditional surveillance systems. By integrating phylogenetic models with ecological and epidemiological data, we can better detect and respond to arbovirus introductions, spillovers and outbreaks that are relevant to both human and veterinary health.sponsorship: S.D. acknowledges support from the Fonds National de la Recherche Scientifique (F.R.S.-FNRS, Belgium; grant no. F.4515.22), from the Research Foundation - Flanders (Fonds voor Wetenschappelijk Onderzoek - Vlaanderen, FWO, Belgium; grant no. G098321N), and from the European Union Horizon 2020 projects MOOD (grant agreement no. 874850) and LEAPS (grant agreement no. 101094685). M.G. acknowledges support from FAPERJ, the National Institutes of Health (NIH), USA (grant no. U01 AI151698) for the United World Arbovirus Research Network (UWARN), and the Novo Nordisk Foundation (grant no. 0094346). N.D.G. is supported by the National Institute of Allergy and Infectious Diseases of the NIH under award no. DP2AI176740. The findings and conclusions in this report are those of the author(s) and do not necessarily represent the official position of the NIH. (Fonds National de la Recherche Scientifique (F.R.S.-FNRS, Belgium)|F.4515.22, Research Foundation - Flanders, FWO, Belgium|G098321N, European Union|874850, European Union|101094685, FAPERJ, National Institutes of Health (NIH), USA|U01 AI151698, Novo Nordisk Foundation|0094346, National Institute of Allergy and Infectious Diseases of the NIH|DP2AI176740)status: Published onlin
Integrated Photonics-Based Focusing through Multimode Fibers
No full textsponsorship: This work was supported by imec's internal funding sources and European Research Council grant no. 805222. (Imec's, European Research Council|805222, European Research Council (ERC)|805222)status: Publishe
Gradient-based optimization of seismic metasurfaces for broadband vibration mitigation in layered soil based on power flow
No full textsponsorship: The results presented in this paper were obtained within the frame of the project G0B8221N "Mitigation of railway induced vibration using seismic metamaterials" funded by the Research Foundation Flanders (FWO) , Belgium. The financial support is gratefully acknowledged. (Research Foundation Flanders (FWO) , Belgium|G0B8221N)status: Accepte
Effect of process parameters on mechanical properties of cold metal transfer manufactured steel wall structures
No full textsponsorship: We would like to gratefully acknowledge Research Foundation Flanders (FWO) for funding of the post-doctoral fellowship "1256522N". We would especially want to thank Burak Karabulut for his guidance and funding support for the research. We would also like to thank Ken Whittaker of Whittaker Engineering for his unrelenting dedication to the accomplishment of our experiments. (Research Foundation Flanders (FWO)|1256522N)status: Publishe
Op AI gebaseerde beslissingsondersteuning voor het monitoren van epilepsie
No full textEpilepsy is one of the most prevalent neurological disorders worldwide, affecting approximately 50 million people. Despite the availability of several treatments, about 30% of patients remain resistant to pharmacological therapy, continuing to experience seizures throughout their lives, which significantly impacts their quality of life. Typically, patients are monitored in specialized Epileptic Monitoring Units (EMUs), where electroencephalography (EEG) is used to measure brain activity and detect seizures. This process is labor-intensive, as the data must be manually reviewed. Outside the hospital, seizure tracking usually relies on self-reported diaries, which are widely recognized as unreliable and inaccurate.
Recent technological advances have enabled the development of wearable devices capable of continuously measuring multiple physiological signals in a non-intrusive manner. However, the longitudinal nature of wearable recordings poses an additional challenge: the large volumes of data generated must be analyzed automatically using artificial intelligence. Although many machine learning (ML) methods have been developed for seizure detection, these methods do not translate directly to wearable data. This is due to electrodes being placed in non-standard scalp locations, lower data quality in ambulatory scenarios, and the presence of noise and artifacts caused by hardware limitations, electrode misplacement, movement, or environmental factors. Therefore, the main objective of this thesis is to design ML frameworks specifically tailored to wearable data and to evaluate their clinical usability as decision-support tools for long-term monitoring of patients with epilepsy.
The thesis is divided into two main parts, addressing focal and generalized absence seizures, as these subtypes present substantially different physiological patterns. All methods are based on data recorded with the Sensor Dot device (Byteflies), which measures behind-the-ear EEG (bte-EEG), electrocardiography (ECG), electromyography (EMG), and movement signals.
The first part focuses on focal seizure detection based on wearable data. It begins with a comparative study of different ML methodologies trained on bte-EEG and ECG, establishing benchmarking frameworks and evaluation methods. Data-centric approaches are then introduced, demonstrating improved model performance and robustness. These frameworks are subsequently evaluated in clinical validation studies. In-hospital results show that algorithm-generated alarms reviewed by clinicians provide favorable clinical utility and detection yields, while home-based monitoring reveals limitations due to artifacts and the challenging nature of capturing ictal patterns generated by focal seizures using bte-EEG electrodes.
The second part adapts and extends the ML frameworks to detect absence seizures, a generalized seizure type characterized by 3 Hz spike-and-wave discharges and transient loss of consciousness. The proposed multimodal pipeline combines bte-EEG-based models with a post-processing stage that incorporates movement data to filter artifacts, reducing false detections. Performance is further enhanced by tailored signal processing techniques, specifically normalization strategies, which reduce variance in noisy bte-EEG data. This results in more compact latent-space representations and improved class separability, thereby increasing robustness in uncontrolled environments. Given that annotated home-based data are scarce in real-world scenarios, this part also introduces a self-supervised learning paradigm that leverages unlabeled home recordings. Using generative approaches, the method embeds domain-specific knowledge in an unsupervised manner. Fine-tuning the pre-trained models with labeled hospital data achieves high sensitivity in detecting seizures.
Overall, this thesis makes important contributions toward the integration of wearable devices into clinical practice. By developing ML frameworks and employing strategies to enhance robustness and performance, it demonstrates the feasibility of reliable seizure monitoring outside specialized hospital environments. These advances not only provide a more accurate alternative to seizure diaries but also reduce the clinical burden while improving monitoring yield, ultimately enabling more effective treatment strategies and paving the way for large-scale longitudinal trials using wearable devices.status: Publishe