<?xml version="1.0" encoding="UTF-8"?>
<rdf:RDF xmlns="http://purl.org/rss/1.0/" xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#" xmlns:dc="http://purl.org/dc/elements/1.1/">
<channel rdf:about="http://hdl.handle.net/10900/42133">
<title>7 Mathematisch-Naturwissenschaftliche Fakultät</title>
<link>http://hdl.handle.net/10900/42133</link>
<description/>
<items>
<rdf:Seq>
<rdf:li rdf:resource="http://hdl.handle.net/10900/183798"/>
<rdf:li rdf:resource="http://hdl.handle.net/10900/183795"/>
<rdf:li rdf:resource="http://hdl.handle.net/10900/183627"/>
<rdf:li rdf:resource="http://hdl.handle.net/10900/183611"/>
</rdf:Seq>
</items>
<dc:date>2026-09-25T21:54:26Z</dc:date>
</channel>
<item rdf:about="http://hdl.handle.net/10900/183798">
<title>Rational Polytopes in Combinatorics and Algebraic Geometry</title>
<link>http://hdl.handle.net/10900/183798</link>
<description>Rational Polytopes in Combinatorics and Algebraic Geometry
Bohnert, Martin
In dieser Dissertation werden rationale Polytope untersucht, wie sie an der Schnittstelle von Kombinatorik und algebraischer Geometrie auftreten. Mithilfe der Gitterbreite erhalten wir zunächst neue Flächenschranken für ebene konvexe Mengen. Im Rahmen der Ehrhart-Theorie klassifizieren wir die Ehrhart-Polynome halbzahliger Polygone. Außerdem untersuchen wir, motiviert durch die birationale Geometrie nicht-degenerierter torischer Hyperflächen, das Fine Interior von Polytopen und geben eine rein kombinatorische Beschreibung von dessen Verhalten unter Dilatationen. Schließlich entwickeln wir effiziente Klassifikationsalgorithmen für einige Fine Interiors sowie für rationale Polygone mit kleinem Nenner und erhalten Verallgemeinerungen der Ungleichung von Scott und der Formel von Pick auf rationale Polygone.; This dissertation investigates rational polytopes in combinatorics and algebraic geometry. Using lattice width data, we first establish new area bounds for planar convex bodies. Within Ehrhart theory, we classify the Ehrhart polynomials of half-integral polygons. Furthermore, we study the Fine interior of polytopes and provide a purely combinatorial description of its behavior under dilations, motivated by the birational geometry of nondegenerate toric hypersurfaces. Finally, we develop efficient classification algorithms for some Fine interiors as well as for rational polygons with small denominators, leading to generalizations of Scott's inequality and Pick's formula to rational polygons.
</description>
<dc:date>2026-09-25T00:00:00Z</dc:date>
</item>
<item rdf:about="http://hdl.handle.net/10900/183795">
<title>Eliciting Code-modulated Visual Evoked Potentials by Non-Recognizable Presentation of Code Sequences</title>
<link>http://hdl.handle.net/10900/183795</link>
<description>Eliciting Code-modulated Visual Evoked Potentials by Non-Recognizable Presentation of Code Sequences
Schlör, Lasse
In den letzten Jahren haben sich Hirn-Computer-Schnittstellen (BCIs), die auf visuell evozierten Potenzialen (VEPs) basieren, als Werkzeug zur Verbesserung der Lebensqualität von Patient:innen mit motorischer Lähmung durch Anwendungen wie blickgesteuerte Tastaturen durchgesetzt. Wahrnehmbares Flimmern zu vermeiden, um Nebenwirkungen wie visuelle Ermüdung oder Krampfanfälle zu mildern, hat sich auf Computerbildschirmen jedoch als Herausforderung herausgestellt. Die vorliegende Arbeit stellt eine neuartige Klasse von Stimulationsmustern vor, die digitale Modulationstechniken verwenden, um wahrnehmbares Flimmern zu minimieren. Die durch diese Muster hervorgerufenen Potenziale, die als codemodulierte visuell evozierte Trägerpotenziale (CCVEPs) bezeichnet werden, sind in der Lage, Fehlerkorrekturcodes (ECCs) zu übertragen. In einem Pilotversuch mit 7 Proband:innen wurden zwei Inkarnationen dieses Paradigmas mittels eines einfachen maschinellen Lernmodells evaluiert, wobei als Modulationsverfahren An-Aus-Umtastung (OOK) bzw. binäre Phasenumtastung (BPSK) verwendet wurden. Eine mittlere Informationsübertragungsrate (ITR) von 69,9 bit/min wurde erreicht und wesentliches Verbesserungspotenzial durch Modifikationen am Modell wird postuliert. Darüber hinaus wird gezeigt, dass Maximalfolgen (MLS)-Kreuzkorrelation auf das CCVEP-Paradigma adaptiert werden kann. 15-klassige MLS-Versatz-Klassifizierung ergab eine mittlere ITR von 124,6 bit/min. Während die im Versuch präsentierten Stimuli den Benutzerkomfort im Vergleich zu herkömmlichen Ansätzen subjektiv erheblich verbessern, ist das Erreichen perfekter Nichterkennbarkeit mit dem CCVEP-Paradigma ein nichttriviales Optimierungsproblem, das weiter untersucht werden sollte.; Recently, brain-computer interfaces (BCIs) based on visual evoked potentials (VEPs) have gained traction as a tool to improve quality of life in motor paralysis patients through applications such as gaze-controlled keyboards. However, avoiding perceptible flickering to mitigate side effects such as visual fatigue or seizures has been challenging on computer screens. The present thesis proposes a novel class of stimulation patterns employing digital modulation techniques to minimize perceptible flickering. The potentials elicited by these patterns, termed code-modulated carrier visual evoked potentials (CCVEPs), are capable of transmitting error correction codes (ECCs). In a pilot experiment with 7 test subjects, two incarnations of this paradigm were evaluated by means of a basic machine learning model, utilizing on-off keying (OOK) and binary phase-shift keying (BPSK) as modulation schemes, respectively. A mean information transfer rate (ITR) of 69.9 bit/min was achieved, and substantial room for improvement from modifications to the model is postulated. In addition, it is demonstrated that maximum length sequence (MLS) cross-correlation can be adapted to the CCVEP paradigm. 15-target MLS shift classification yielded a mean ITR of 124.6 bit/min. While the stimuli presented in the experiment subjectively improve user comfort vastly compared to traditional approaches, achieving perfect non-recognizability with the CCVEP paradigm is a nontrivial optimization problem worthy of further study.
</description>
<dc:date>2021-11-24T00:00:00Z</dc:date>
</item>
<item rdf:about="http://hdl.handle.net/10900/183627">
<title>Optimization, Functionalization and Application of Nanobodies against Miro1 and Drp1 to study Mitochondrial Dynamics</title>
<link>http://hdl.handle.net/10900/183627</link>
<description>Optimization, Functionalization and Application of Nanobodies against Miro1 and Drp1 to study Mitochondrial Dynamics
Fröhlich, Theresa Marie
To fulfil their multiple cellular functions, mitochondria constantly adapt to changing requirements by fusion, fission and active transport within the cell. These processes are coordinated by two key proteins: Drp1, which is responsible for the membrane constriction during mitochondrial fission, and Miro1, which connects the mitochondria to motor proteins, thus enabling mitochondrial movement.&#13;
Drp1 dysregulation was linked to multiple diseases including Parkinson’s and cardiovascular diseases thus making Drp1 an attractive biomarker and a potential therapeutic target. To leverage this potential, a deeper understanding of Drp1 and Miro1 is required, which is currently impeded by the available research tools. To facilitate biochemical and cell biological studies of endogenous Drp1, Drp1-specific nanobodies (Nbs) previously identified by a phage display screen were formatted into versatile research tools. Functionalized into capture matrices, the Drp1-Nbs allowed the enrichment of endogenous Drp1 and thereby the identification of Drp1 interaction partners by proteomic analysis. Furthermore, a fluorophore-labelled bivalent Drp1-Nb (bivD7AF647) was generated and successfully applied for super-resolution STORM microscopy of endogenous Drp1. For real-time imaging of Drp1, Nb-fluorescent protein fusions, so-called chromobodies (Cbs), were intracellularly expressed in mammalian cells and the signal-to-noise ratio further improved by generating “turnover-accelerated” Cbs. With these probes, the dynamic relocalization of endogenous Drp1 upon compound-mediated induction of fission was visualized in living cells. Additionally, the intracellularly functional Drp1-Nbs were engineered into inducible TRIM21-based degrons for targeted depletion of Drp1 in live cells.&#13;
For proteomic studies of endogenous Miro1, a previously generated Miro1-specific Nb was formatted into a bivalent affinity matrix, which was successfully applied to analyse the Miro1 interactome.&#13;
To establish an approach for bringing fluorescently labelled Nbs into live cells, a recently described cholesterol-coomassie based approach for transducing proteins in live cells was tested for different Nbs. Although successful transduction of the Nbs into the cytosol was observed in principle, the method suffered from high cellular toxicity and low efficiency.&#13;
Due to their versatility, the Nb-based research tools developed and validated in this study can facilitate further investigations into Drp1 and Miro1 and thus help answer open questions regarding their role in mitochondrial dynamics and pathophysiology.
</description>
<dc:date>2026-09-23T00:00:00Z</dc:date>
</item>
<item rdf:about="http://hdl.handle.net/10900/183611">
<title>The Construction of Algorithmic Predictions in Society</title>
<link>http://hdl.handle.net/10900/183611</link>
<description>The Construction of Algorithmic Predictions in Society
Höltgen, Benedikt
Immer mehr Entscheidungen in gesellschaftlichen Kontexten werden automatisiert.&#13;
Algorithmische Vorhersagen werden dabei häufig so interpretiert, dass sie Zusammenhänge zwischen realen Aspekten der Welt identifizieren (oder zumindest approximieren), die unabhängig von den gewählten technischen Methoden sind. Diese Annahme prägt sowohl die Fachliteratur als auch die öffentliche Debatte und wird oft als Rechtfertigung von Entscheidungen herangezogen, die auf solchen Vorhersagen basieren. &#13;
Wie ich argumentiere, beruht dieses Verständnis statistischer Regelmäßigkeiten auf (i) einem von Naturgesetzen geprägten Wissenschaftsverständnis, (ii) einem „Casino-Verständnis“ von Wahrscheinlichkeiten als objektiven Größen sowie (iii) der zentralen Rolle, die Gesetzmäßigkeiten in der Statistik zugesprochen wird.&#13;
&#13;
Im ersten Teil dieser Arbeit zeige ich, dass Wahrscheinlichkeiten generell konstruiert sind. Dies wird oft übersehen, da sie nicht direkt evaluierbar sind. Ich systematisiere die Evaluierung probabilistischer Vorhersagen mittels Kalibrierung und vereinheitliche unterschiedliche Auffassungen von Wahrscheinlichkeiten über Gemeinsamkeiten in der Abhängigkeit von Modellen.&#13;
&#13;
Im zweiten Teil entwickle ich ein Verständnis von maschinellem Lernen für gesellschaftliche Kontexte, welches ohne die Annahme von datengenerierenden Wahrscheinlichkeitsverteilungen auskommt, und zeige, dass diese Annahme in der Praxis irreführend ist.&#13;
Ich widerlege die daraus abgeleitete Vorstellung, dass eine Berücksichtigung von Fairness notwendigerweise mit schlechteren Vorhersagen einhergehen muss, sowohl theoretisch als auch empirisch. Zudem zeige ich, dass---entgegen der gängigen Auffassung---die Einbeziehung demografischer Merkmale Ungleichheiten und Stereotype verstärken kann.&#13;
&#13;
Im dritten Teil erweitere ich den Anwendungsbereich der vorangegangenen Überlegungen. Ich entwickle ein mathematisches Modell für kausale Inferenz, dessen Annahmen und Vorhersagen leichter überprüfbar sind, und schließe mit einer Kritik verbreiteter Auffassungen über künstliche Intelligenz.&#13;
&#13;
Zusammengefasst lege ich Probleme des vorherrschenden Verständnisses algorithmischer Vorhersagen dar und leiste sowohl konzeptionelle als auch technische Beiträge, die den konstruierten Charakter von Vorhersagen aufzeigen und die theoretische Analyse näher an die soziale Realität heranführen.; As data-driven decision-making is increasingly pervasive in society, algorithmic&#13;
predictions are often understood as identifying, or at least approximating, prop-&#13;
erties of the world that are independent of the technical tools; this applies to&#13;
the technical literature as much as to the public. Individual decisions based&#13;
on such predictions are then justified by the claim that the models track real&#13;
relationships between measured attributes. I argue that this view of statistical&#13;
regularities is based on (i) the common understanding of science as identifying&#13;
natural laws, (ii) the casino understanding of chances or risks as something that&#13;
can be identified, and (iii) the fundamental role of statistical regularities in the&#13;
standard mathematical framework of statistics and machine learning.&#13;
In the first part of this thesis, I demonstrate that probabilities are generally&#13;
constructed and that contrary intuitions rest at least partly on the fact that we&#13;
cannot directly evaluate them. In particular, I systematise the evaluation of prob-&#13;
abilistic predictions through notions of calibration and unify different perspect-&#13;
ives on probability by emphasising the ubiquity of model-dependence.&#13;
In the second part, I present a constructive perspective on machine learn-&#13;
ing for societal contexts in particular. I show that the assumption of a data-&#13;
generating distribution is not necessary for modelling machine learning and&#13;
that it can be misleading in practice. Theoretically and empirically, I disprove&#13;
the idea derived from this assumption that fairness considerations necessarily&#13;
require a deviation from better predictions. I also show that (against received&#13;
wisdom) the inclusion of demographic attributes can exacerbate inequality and&#13;
that their current use can be harmful even in auditing for racial discrimination.&#13;
In the third and last part, I extend the scope of the foregoing considerations&#13;
beyond the supervised learning setting. For causal inference tasks, I present a&#13;
framework dispensing with data-generating distributions, which not only cap-&#13;
tures established estimators but also paves the way for greater reliability and&#13;
accountability by making the required assumptions testable. I conclude with a&#13;
more general critique of common understandings of artificial intelligence.&#13;
Taken together, I demonstrate problems with the prevalent understanding of&#13;
algorithmic predictions and offer conceptual and technical contributions that, in&#13;
various ways, emphasise their constructed nature and bring theoretical analysis&#13;
closer to social reality.
</description>
<dc:date>2026-09-22T00:00:00Z</dc:date>
</item>
</rdf:RDF>
