@article{HerssensCowburnAlbrachtetal.2022, author = {Herssens, Nolan and Cowburn, James and Albracht, Kirsten and Braunstein, Bjoern and Cazzola, Dario and Colyer, Steffi and Minetti, Alberto E. and Pavei, Gaspare and Rittweger, J{\"o}rn and Weber, Tobias and Green, David A.}, title = {Movement in low gravity environments (MoLo) programme - the MoLo-L.O.O.P. study protocol}, series = {PLOS ONE / Public Library of Science}, volume = {17}, journal = {PLOS ONE / Public Library of Science}, number = {11}, editor = {Cattaneo, Luigi}, publisher = {Plos}, address = {San Francisco}, issn = {1932-6203}, doi = {10.1371/journal.pone.0278051}, pages = {e0278051}, year = {2022}, abstract = {Exposure to prolonged periods in microgravity is associated with deconditioning of the musculoskeletal system due to chronic changes in mechanical stimulation. Given astronauts will operate on the Lunar surface for extended periods of time, it is critical to quantify both external (e.g., ground reaction forces) and internal (e.g., joint reaction forces) loads of relevant movements performed during Lunar missions. Such knowledge is key to predict musculoskeletal deconditioning and determine appropriate exercise countermeasures associated with extended exposure to hypogravity.}, language = {en} } @unpublished{BornheimNiklasBlanecketal.2023, author = {Bornheim, Tobias and Niklas, Grieger and Blaneck, Patrick Gustav and Bialonski, Stephan}, title = {Preprint: Speaker attribution in German parliamentary debates with QLoRA-adapted large language models}, series = {Journal for Language Technology and Computational Linguistics}, journal = {Journal for Language Technology and Computational Linguistics}, doi = {10.48550/arXiv.2309.09902}, pages = {8 Seiten}, year = {2023}, abstract = {The growing body of political texts opens up new opportunities for rich insights into political dynamics and ideologies but also increases the workload for manual analysis. Automated speaker attribution, which detects who said what to whom in a speech event and is closely related to semantic role labeling, is an important processing step for computational text analysis. We study the potential of the large language model family Llama 2 to automate speaker attribution in German parliamentary debates from 2017-2021. We fine-tune Llama 2 with QLoRA, an efficient training strategy, and observe our approach to achieve competitive performance in the GermEval 2023 Shared Task On Speaker Attribution in German News Articles and Parliamentary Debates. Our results shed light on the capabilities of large language models in automating speaker attribution, revealing a promising avenue for computational analysis of political discourse and the development of semantic role labeling systems.}, language = {en} } @article{WendlandtKochBritzetal.2023, author = {Wendlandt, Tim and Koch, Claudia and Britz, Beate and Liedek, Anke and Schmidt, Nora and Werner, Stefan and Gleba, Yuri and Vahidpour, Farnoosh and Welden, Melanie and Poghossian, Arshak and Sch{\"o}ning, Michael Josef}, title = {Facile Purification and Use of Tobamoviral Nanocarriers for Antibody-Mediated Display of a Two-Enzyme System}, series = {Viruses}, volume = {9}, journal = {Viruses}, number = {15}, publisher = {MDPI}, address = {Basel}, issn = {1999-4915}, doi = {doi.org/10.3390/v15091951}, pages = {Artikel 1951}, year = {2023}, abstract = {Immunosorbent turnip vein clearing virus (TVCV) particles displaying the IgG-binding domains D and E of Staphylococcus aureus protein A (PA) on every coat protein (CP) subunit (TVCVPA) were purified from plants via optimized and new protocols. The latter used polyethylene glycol (PEG) raw precipitates, from which virions were selectively re-solubilized in reverse PEG concentration gradients. This procedure improved the integrity of both TVCVPA and the wild-type subgroup 3 tobamovirus. TVCVPA could be loaded with more than 500 IgGs per virion, which mediated the immunocapture of fluorescent dyes, GFP, and active enzymes. Bi-enzyme ensembles of cooperating glucose oxidase and horseradish peroxidase were tethered together on the TVCVPA carriers via a single antibody type, with one enzyme conjugated chemically to its Fc region, and the other one bound as a target, yielding synthetic multi-enzyme complexes. In microtiter plates, the TVCVPA-displayed sugar-sensing system possessed a considerably increased reusability upon repeated testing, compared to the IgG-bound enzyme pair in the absence of the virus. A high coverage of the viral adapters was also achieved on Ta2O5 sensor chip surfaces coated with a polyelectrolyte interlayer, as a prerequisite for durable TVCVPA-assisted electrochemical biosensing via modularly IgG-assembled sensor enzymes.}, language = {en} } @article{BertzMolinnusSchoeningetal.2023, author = {Bertz, Morten and Molinnus, Denise and Sch{\"o}ning, Michael Josef and Homma, Takayuki}, title = {Real-time monitoring of H₂O₂ sterilization on individual bacillus atrophaeus spores by optical sensing with trapping Raman spectroscopy}, series = {Chemosensors}, volume = {8}, journal = {Chemosensors}, number = {11}, publisher = {MDPI}, address = {Basel}, issn = {2227-9040}, doi = {10.3390/chemosensors11080445}, pages = {Artikel 445}, year = {2023}, abstract = {Hydrogen peroxide (H₂O₂), a strong oxidizer, is a commonly used sterilization agent employed during aseptic food processing and medical applications. To assess the sterilization efficiency with H₂O₂, bacterial spores are common microbial systems due to their remarkable robustness against a wide variety of decontamination strategies. Despite their widespread use, there is, however, only little information about the detailed time-resolved mechanism underlying the oxidative spore death by H₂O₂. In this work, we investigate chemical and morphological changes of individual Bacillus atrophaeus spores undergoing oxidative damage using optical sensing with trapping Raman microscopy in real-time. The time-resolved experiments reveal that spore death involves two distinct phases: (i) an initial phase dominated by the fast release of dipicolinic acid (DPA), a major spore biomarker, which indicates the rupture of the spore's core; and (ii) the oxidation of the remaining spore material resulting in the subsequent fragmentation of the spores' coat. Simultaneous observation of the spore morphology by optical microscopy corroborates these mechanisms. The dependence of the onset of DPA release and the time constant of spore fragmentation on H₂O₂ shows that the formation of reactive oxygen species from H₂O₂ is the rate-limiting factor of oxidative spore death.}, language = {en} } @phdthesis{Gaigall2023, author = {Gaigall, Daniel}, title = {On selected problems in multivariate analysis}, doi = {10.15488/14304}, pages = {17 Seiten}, year = {2023}, abstract = {Selected problems in the field of multivariate statistical analysis are treated. Thereby, one focus is on the paired sample case. Among other things, statistical testing problems of marginal homogeneity are under consideration. In detail, properties of Hotelling's T² test in a special parametric situation are obtained. Moreover, the nonparametric problem of marginal homogeneity is discussed on the basis of possibly incomplete data. In the bivariate data case, properties of the Hoeffding-Blum-Kiefer-Rosenblatt independence test statistic on the basis of partly not identically distributed data are investigated. Similar testing problems are treated within the scope of the application of a result for the empirical process of the concomitants for partly categorial data. Furthermore, testing changes in the modeled solvency capital requirement of an insurance company by means of a paired sample from an internal risk model is discussed. Beyond the paired sample case, a new asymptotic relative efficiency concept based on the expected volumes of multidimensional confidence regions is introduced. Besides, a new approach for the treatment of the multi-sample goodness-of-fit problem is presented. Finally, a consistent test for the treatment of the goodness-of-fit problem is developed for the background of huge or infinite dimensional data.}, language = {en} } @article{MoraisSumanSchoeningetal.2023, author = {Morais, Paulo V. and Suman, Pedro H. and Sch{\"o}ning, Michael Josef and Siqueira Junior, Jos{\´e} R. and Orlandi, Marcelo O.}, title = {Layer-by-layer film based on Sn₃O₄ nanobelts as sensing units to detect heavy metals using a capacitive field-effect sensor platform}, series = {Chemosensors}, volume = {11}, journal = {Chemosensors}, number = {8}, publisher = {MDPI}, address = {Basel}, issn = {2227-9040}, doi = {10.3390/chemosensors11080436}, pages = {Artikel 436}, year = {2023}, abstract = {Lead and nickel, as heavy metals, are still used in industrial processes, and are classified as "environmental health hazards" due to their toxicity and polluting potential. The detection of heavy metals can prevent environmental pollution at toxic levels that are critical to human health. In this sense, the electrolyte-insulator-semiconductor (EIS) field-effect sensor is an attractive sensing platform concerning the fabrication of reusable and robust sensors to detect such substances. This study is aimed to fabricate a sensing unit on an EIS device based on Sn₃O₄ nanobelts embedded in a polyelectrolyte matrix of polyvinylpyrrolidone (PVP) and polyacrylic acid (PAA) using the layer-by-layer (LbL) technique. The EIS-Sn₃O₄ sensor exhibited enhanced electrochemical performance for detecting Pb²⁺ and Ni²⁺ ions, revealing a higher affinity for Pb²⁺ ions, with sensitivities of ca. 25.8 mV/decade and 2.4 mV/decade, respectively. Such results indicate that Sn₃O₄ nanobelts can contemplate a feasible proof-of-concept capacitive field-effect sensor for heavy metal detection, envisaging other future studies focusing on environmental monitoring.}, language = {en} } @article{KuchlerGuenthnerRibeiroetal.2023, author = {Kuchler, Timon and G{\"u}nthner, Roman and Ribeiro, Andrea and Hausinger, Renate and Streese, Lukas and W{\"o}hnl, Anna and Kesseler, Veronika and Negele, Johanna and Assali, Tarek and Carbajo-Lozoya, Javier and Lech, Maciej and Adorjan, Kristina and Stubbe, Hans Christian and Hanssen, Henner and Kotliar, Konstantin and Haller, Berhard and Heemann, Uwe and Schmaderer, Christoph}, title = {Persistent endothelial dysfunction in post-COVID-19 syndrome and its associations with symptom severity and chronic inflammation}, volume = {26}, publisher = {Springer Nature}, address = {Dordrecht}, doi = {10.1007/s10456-023-09885-6}, pages = {547 -- 563}, year = {2023}, abstract = {Background Post-COVID-19 syndrome (PCS) is a lingering disease with ongoing symptoms such as fatigue and cognitive impairment resulting in a high impact on the daily life of patients. Understanding the pathophysiology of PCS is a public health priority, as it still poses a diagnostic and treatment challenge for physicians. Methods In this prospective observational cohort study, we analyzed the retinal microcirculation using Retinal Vessel Analysis (RVA) in a cohort of patients with PCS and compared it to an age- and gender-matched healthy cohort (n = 41, matched out of n = 204). Measurements and main results PCS patients exhibit persistent endothelial dysfunction (ED), as indicated by significantly lower venular flicker-induced dilation (vFID; 3.42\% ± 1.77\% vs. 4.64\% ± 2.59\%; p = 0.02), narrower central retinal artery equivalent (CRAE; 178.1 [167.5-190.2] vs. 189.1 [179.4-197.2], p = 0.01) and lower arteriolar-venular ratio (AVR; (0.84 [0.8-0.9] vs. 0.88 [0.8-0.9], p = 0.007). When combining AVR and vFID, predicted scores reached good ability to discriminate groups (area under the curve: 0.75). Higher PCS severity scores correlated with lower AVR (R = - 0.37 p = 0.017). The association of microvascular changes with PCS severity were amplified in PCS patients exhibiting higher levels of inflammatory parameters. Conclusion Our results demonstrate that prolonged endothelial dysfunction is a hallmark of PCS, and impairments of the microcirculation seem to explain ongoing symptoms in patients. As potential therapies for PCS emerge, RVA parameters may become relevant as clinical biomarkers for diagnosis and therapy management.}, language = {en} } @inproceedings{KloeserBuesgenKohletal.2023, author = {Kl{\"o}ser, Lars and B{\"u}sgen, Andr{\´e} and Kohl, Philipp and Kraft, Bodo and Z{\"u}ndorf, Albert}, title = {Explaining relation classification models with semantic extents}, series = {DeLTA 2023: Deep Learning Theory and Applications}, booktitle = {DeLTA 2023: Deep Learning Theory and Applications}, editor = {Conte, Donatello and Fred, Ana and Gusikhin, Oleg and Sansone, Carlo}, publisher = {Springer}, address = {Cham}, isbn = {978-3-031-39058-6 (Print)}, doi = {10.1007/978-3-031-39059-3_13}, pages = {189 -- 208}, year = {2023}, abstract = {In recent years, the development of large pretrained language models, such as BERT and GPT, significantly improved information extraction systems on various tasks, including relation classification. State-of-the-art systems are highly accurate on scientific benchmarks. A lack of explainability is currently a complicating factor in many real-world applications. Comprehensible systems are necessary to prevent biased, counterintuitive, or harmful decisions. We introduce semantic extents, a concept to analyze decision patterns for the relation classification task. Semantic extents are the most influential parts of texts concerning classification decisions. Our definition allows similar procedures to determine semantic extents for humans and models. We provide an annotation tool and a software framework to determine semantic extents for humans and models conveniently and reproducibly. Comparing both reveals that models tend to learn shortcut patterns from data. These patterns are hard to detect with current interpretability methods, such as input reductions. Our approach can help detect and eliminate spurious decision patterns during model development. Semantic extents can increase the reliability and security of natural language processing systems. Semantic extents are an essential step in enabling applications in critical areas like healthcare or finance. Moreover, our work opens new research directions for developing methods to explain deep learning models.}, language = {en} } @inproceedings{KohlFreyerKraemeretal.2023, author = {Kohl, Philipp and Freyer, Nils and Kr{\"a}mer, Yoka and Werth, Henri and Wolf, Steffen and Kraft, Bodo and Meinecke, Matthias and Z{\"u}ndorf, Albert}, title = {ALE: a simulation-based active learning evaluation framework for the parameter-driven comparison of query strategies for NLP}, series = {Deep Learning Theory and Applications. DeLTA 2023. Communications in Computer and Information Science}, booktitle = {Deep Learning Theory and Applications. DeLTA 2023. Communications in Computer and Information Science}, editor = {Conte, Donatello and Fred, Ana and Gusikhin, Oleg and Sansone, Carlo}, publisher = {Springer}, address = {Cham}, isbn = {978-3-031-39058-6 (Print)}, doi = {978-3-031-39059-3}, pages = {235 -- 253}, year = {2023}, abstract = {Supervised machine learning and deep learning require a large amount of labeled data, which data scientists obtain in a manual, and time-consuming annotation process. To mitigate this challenge, Active Learning (AL) proposes promising data points to annotators they annotate next instead of a subsequent or random sample. This method is supposed to save annotation effort while maintaining model performance. However, practitioners face many AL strategies for different tasks and need an empirical basis to choose between them. Surveys categorize AL strategies into taxonomies without performance indications. Presentations of novel AL strategies compare the performance to a small subset of strategies. Our contribution addresses the empirical basis by introducing a reproducible active learning evaluation (ALE) framework for the comparative evaluation of AL strategies in NLP. The framework allows the implementation of AL strategies with low effort and a fair data-driven comparison through defining and tracking experiment parameters (e.g., initial dataset size, number of data points per query step, and the budget). ALE helps practitioners to make more informed decisions, and researchers can focus on developing new, effective AL strategies and deriving best practices for specific use cases. With best practices, practitioners can lower their annotation costs. We present a case study to illustrate how to use the framework.}, language = {en} } @inproceedings{BuesgenKloeserKohletal.2023, author = {B{\"u}sgen, Andr{\´e} and Kl{\"o}ser, Lars and Kohl, Philipp and Schmidts, Oliver and Kraft, Bodo and Z{\"u}ndorf, Albert}, title = {From cracked accounts to fake IDs: user profiling on German telegram black market channels}, series = {Data Management Technologies and Applications}, booktitle = {Data Management Technologies and Applications}, editor = {Cuzzocrea, Alfredo and Gusikhin, Oleg and Hammoudi, Slimane and Quix, Christoph}, publisher = {Springer}, address = {Cham}, isbn = {978-3-031-37889-8 (Print)}, doi = {10.1007/978-3-031-37890-4_9}, pages = {176 -- 202}, year = {2023}, abstract = {Messenger apps like WhatsApp and Telegram are frequently used for everyday communication, but they can also be utilized as a platform for illegal activity. Telegram allows public groups with up to 200.000 participants. Criminals use these public groups for trading illegal commodities and services, which becomes a concern for law enforcement agencies, who manually monitor suspicious activity in these chat rooms. This research demonstrates how natural language processing (NLP) can assist in analyzing these chat rooms, providing an explorative overview of the domain and facilitating purposeful analyses of user behavior. We provide a publicly available corpus of annotated text messages with entities and relations from four self-proclaimed black market chat rooms. Our pipeline approach aggregates the extracted product attributes from user messages to profiles and uses these with their sold products as features for clustering. The extracted structured information is the foundation for further data exploration, such as identifying the top vendors or fine-granular price analyses. Our evaluation shows that pretrained word vectors perform better for unsupervised clustering than state-of-the-art transformer models, while the latter is still superior for sequence labeling.}, language = {en} }