004 Datenverarbeitung; Informatik
Refine
Year of publication
Document Type
- Article (259)
- Doctoral Thesis (147)
- Working Paper (122)
- Bachelor Thesis (53)
- Conference Proceeding (53)
- Diploma Thesis (48)
- Preprint (44)
- Part of a Book (42)
- Contribution to a Periodical (39)
- diplomthesis (31)
Is part of the Bibliography
- no (905)
Keywords
- Lambda-Kalkül (21)
- Inklusion (13)
- Formale Semantik (11)
- Barrierefreiheit (10)
- Digitalisierung (10)
- Operationale Semantik (9)
- artificial intelligence (9)
- data science (9)
- lambda calculus (9)
- machine learning (9)
Institute
- Informatik (475)
- Informatik und Mathematik (104)
- Präsidium (74)
- Frankfurt Institute for Advanced Studies (FIAS) (54)
- Medizin (54)
- Wirtschaftswissenschaften (45)
- Physik (35)
- Hochschulrechenzentrum (24)
- studiumdigitale (24)
- Extern (12)
- Mathematik (11)
- Rechtswissenschaft (11)
- Biowissenschaften (10)
- Geschichtswissenschaften (8)
- Universitätsbibliothek (8)
- Center for Scientific Computing (CSC) (7)
- Deutsches Institut für Internationale Pädagogische Forschung (DIPF) (7)
- Gesellschaftswissenschaften (7)
- Psychologie (7)
- Neuere Philologien (6)
- Ernst Strüngmann Institut (5)
- Goethe-Zentrum für Wissenschaftliches Rechnen (G-CSC) (5)
- Erziehungswissenschaften (4)
- Geowissenschaften (4)
- Kulturwissenschaften (4)
- MPI für Hirnforschung (4)
- Exzellenzcluster Die Herausbildung normativer Ordnungen (3)
- House of Finance (HoF) (3)
- Senckenbergische Naturforschende Gesellschaft (3)
- Sustainable Architecture for Finance in Europe (SAFE) (3)
- Biochemie und Chemie (2)
- Center for Financial Studies (CFS) (2)
- Geographie (2)
- Psychologie und Sportwissenschaften (2)
- Sprach- und Kulturwissenschaften (2)
- Biochemie, Chemie und Pharmazie (1)
- Biodiversität und Klima Forschungszentrum (BiK-F) (1)
- Buchmann Institut für Molekulare Lebenswissenschaften (BMLS) (1)
- E-Finance Lab e.V. (1)
- Geowissenschaften / Geographie (1)
- Institut für Ökologie, Evolution und Diversität (1)
- Interdisziplinäres Zentrum für Neurowissenschaften Frankfurt (IZNF) (1)
- MPI für empirische Ästhetik (1)
- Schreibzentrum (1)
- keine Angabe Institut (1)
»Just Computing?« – der Titel des Auftakts war bewusst doppeldeutig gewählt. Bedeutet die digitale Transformation »nur«, dass rechnergestützte Methoden Prozesse allerorten optimieren? Oder ist damit auch ein Wandel gemeint, der »gerecht« gestaltet werden kann? Antworten auf solch grundlegende Fragen – so viel stellte Uni-Präsident Enrico Schleiff bei der Eröffnung klar – können nur gemeinsam erarbeitet werden mit einem neuen, Fachgrenzen überschreitenden Denken. »Eine Mammutaufgabe und nicht ohne Risiko«, so Schleiff, aber notwendig, weshalb er das Zentrum in seiner Präsidentschaftsbewerbung auch »versprochen« habe. Nun ist es da, und zwölf neue »ausfinanzierte« Professuren sollen in den kommenden zwei Jahren dorthin berufen werden, gab der Sprecher des Gründungsvorstands Christoph Burchard bekannt. Neben dem Juristen Burchard gehören diesem Gründungsvorstand die Bioinformatikerin Franziska Matthäus an, der Informatiker Ulrich Meyer und die Erziehungswissenschaftlerin Juliane Engel – die angestrebte Perspektivenvielfalt ist auf dieser Ebene also schon einmal vorhanden.
DNA points accumulation for imaging in nanoscale topography (DNA-PAINT) is a super-resolution technique with relatively easy-to-implement multi-target imaging. However, image acquisition is slow as sufficient statistical data has to be generated from spatio-temporally isolated single emitters. Here, we trained the neural network (NN) DeepSTORM to predict fluorophore positions from high emitter density DNA-PAINT data. This achieves image acquisition in one minute. We demonstrate multi-color super-resolution imaging of structure-conserved semi-thin neuronal tissue and imaging of large samples. This improvement can be integrated into any single-molecule microscope and enables fast single-molecule super-resolution microscopy.
Electrocardiograms (ECG) record the heart activity and are the most common and reliable method to detect cardiac arrhythmias, such as atrial fibrillation (AFib). Lately, many commercially available devices such as smartwatches are offering ECG monitoring. Therefore, there is increasing demand for designing deep learning models with the perspective to be physically implemented on these small portable devices with limited energy supply. In this paper, a workflow for the design of small, energy-efficient recurrent convolutional neural network (RCNN) architecture for AFib detection is proposed. However, the approach can be well generalized to every type of long time series. In contrast to previous studies, that demand thousands of additional network neurons and millions of extra model parameters, the logical steps for the generation of a CNN with only 114 trainable parameters are described. The model consists of a small segmented CNN in combination with an optimal energy classifier. The architectural decisions are made by using the energy consumption as a metric in an equally important way as the accuracy. The optimization steps are focused on the software which can be embedded afterwards on a physical chip. Finally, a comparison with some previous relevant studies suggests that the widely used huge CNNs for similar tasks are mostly redundant and unessentially computationally expensive.
Electrocardiograms (ECG) record the heart activity and are the most common and reliable method to detect cardiac arrhythmias, such as atrial fibrillation (AFib). Lately, many commercially available devices such as smartwatches are offering ECG monitoring. Therefore, there is increasing demand for designing deep learning models with the perspective to be physically implemented on these small portable devices with limited energy supply. In this paper, a workflow for the design of small, energy-efficient recurrent convolutional neural network (RCNN) architecture for AFib detection is proposed. However, the approach can be well generalized to every type of long time series. In contrast to previous studies, that demand thousands of additional network neurons and millions of extra model parameters, the logical steps for the generation of a CNN with only 114 trainable parameters are described. The model consists of a small segmented CNN in combination with an optimal energy classifier. The architectural decisions are made by using the energy consumption as a metric in an equally important way as the accuracy. The optimisation steps are focused on the software which can be embedded afterwards on a physical chip. Finally, a comparison with some previous relevant studies suggests that the widely used huge CNNs for similar tasks are mostly redundant and unessentially computationally expensive.
Non-coding variations located within regulatory elements may alter gene expression by modifying Transcription Factor (TF) binding sites and thereby lead to functional consequences like various traits or diseases. To understand these molecular mechanisms, different TF models are being used to assess the effect of DNA sequence variations, such as Single Nucleotide Polymorphisms (SNPs). However, few statistical approaches exist to compute statistical significance of results but they often are slow for large sets of SNPs, such as data obtained from a genome-wide association study (GWAS) or allele-specific analysis of chromatin data.
Results We investigate the distribution of maximal differential TF binding scores for general computational models that assess TF binding. We find that a modified Laplace distribution can adequately approximate the empirical distributions. A benchmark on in vitro and in vivo data sets showed that our new approach improves on an existing method in terms of performance and speed. In applications on large sets of eQTL and GWAS SNPs we could illustrate the usefulness of the novel statistic to highlight cell type specific regulators and TF target genes.
Conclusions Our approach allows the evaluation of DNA changes that induce differential TF binding in a fast and accurate manner, permitting computations on large mutation data sets. An implementation of the novel approach is freely available at https://github.com/SchulzLab/SNEEP.
A key competence for open-ended learning is the formation of increasingly abstract representations useful for driving complex behavior. Abstract representations ignore specific details and facilitate generalization. Here we consider the learning of abstract representations in a multi-modal setting with two or more input modalities. We treat the problem as a lossy compression problem and show that generic lossy compression of multimodal sensory input naturally extracts abstract representations that tend to strip away modalitiy specific details and preferentially retain information that is shared across the different modalities. Furthermore, we propose an architecture to learn abstract representations by identifying and retaining only the information that is shared across multiple modalities while discarding any modality specific information.
Recent advances in artificial neural networks enabled the quick development of new learning algorithms, which, among other things, pave the way to novel robotic applications. Traditionally, robots are programmed by human experts so as to accomplish pre-defined tasks. Such robots must operate in a controlled environment to guarantee repeatability, are designed to solve one unique task and require costly hours of development. In developmental robotics, researchers try to artificially imitate the way living beings acquire their behavior by learning. Learning algorithms are key to conceive versatile and robust robots that can adapt to their environment and solve multiple tasks efficiently. In particular, Reinforcement Learning (RL) studies the acquisition of skills through teaching via rewards. In this thesis, we will introduce RL and present recent advances in RL applied to robotics. We will review Intrinsically Motivated (IM) learning, a special form of RL, and we will apply in particular the Active Efficient Coding (AEC) principle to the learning of active vision. We also propose an overview of Hierarchical Reinforcement Learning (HRL), an other special form of RL, and apply its principle to a robotic manipulation task.
This thesis presents a first-of-its-kind phenomenological framework that formally describes the development of acquired epilepsy and the role of the neuro-immune axis in this development. Formulated as a system of nonlinear differential equations, the model describes the interaction of processes such as neuroinflammation, blood- brain barrier disruption, neuronal death, circuit remodeling, and epileptic seizures. The model allows for the simulation of epilepsy development courses caused by a variety of neurological injuries. The simulation results are in agreement with ex- perimental findings from three distinct animal models of epileptogenesis. Simula- tions capture injury-specific temporal patterns of seizure occurrence, neuroinflam- mation, blood-brain barrier leakage, and progression of neuronal death. In addition, the model provides insights into phenomena related to epileptogenesis such as the emergence of paradoxically long time scales of disease development after injury, the dose-dependence of epileptogenesis features on injury severity, and the variability of clinical outcomes in subjects exposed to identical injury. Moreover, the developed framework allows for the simulation of therapeutic interventions, which provides insights into the injury-specificity of prominent intervention strategies. Thus, the model can be used as an in silico tool for the generation of testable predictions, which may aid pre-clinical research for the development of epilepsy treatments.
The cortical networks that underlie behavior exhibit an orderly functional organization at local and global scales, which is readily evident in the visual cortex of carnivores and primates1-6. Here, neighboring columns of neurons represent the full range of stimulus orientations and contribute to distributed networks spanning several millimeters2,7-11. However, the principles governing functional interactions that bridge this fine-scale functional architecture and distant network elements are unclear, and the emergence of these network interactions during development remains unexplored. Here, by using in vivo wide-field and 2-photon calcium imaging of spontaneous activity patterns in mature ferret visual cortex, we find widespread and specific modular correlation patterns that accurately predict the local structure of visually-evoked orientation columns from the spontaneous activity of neurons that lie several millimeters away. The large-scale networks revealed by correlated spontaneous activity show abrupt ‘fractures’ in continuity that are in tight register with evoked orientation pinwheels. Chronic in vivo imaging demonstrates that these large-scale modular correlation patterns and fractures are already present at early stages of cortical development and predictive of the mature network structure. Silencing feed-forward drive through either retinal or thalamic blockade does not affect network structure suggesting a cortical origin for this large-scale correlated activity, despite the immaturity of long-range horizontal network connections in the early cortex. Using a circuit model containing only local connections, we demonstrate that such a circuit is sufficient to generate large-scale correlated activity, while also producing correlated networks showing strong fractures, a reduced dimensionality, and an elongated local correlation structure, all in close agreement with our empirical data. These results demonstrate the precise local and global organization of cortical networks revealed through correlated spontaneous activity and suggest that local connections in early cortical circuits may generate structured long-range network correlations that underlie the subsequent formation of visually-evoked distributed functional networks.
The fundamental structure of cortical networks arises early in development prior to the onset of sensory experience. However, how endogenously generated networks respond to the onset of sensory experience, and how they form mature sensory representations with experience remains unclear. Here we examine this ‘nature-nurture transform’ using in vivo calcium imaging in ferret visual cortex. At eye-opening, visual stimulation evokes robust patterns of cortical activity that are highly variable within and across trials, severely limiting stimulus discriminability. Initial evoked responses are distinct from spontaneous activity of the endogenous network. Visual experience drives the development of low-dimensional, reliable representations aligned with spontaneous activity. A computational model shows that alignment of novel visual inputs and recurrent cortical networks can account for the emergence of reliable visual representations.
The fundamental structure of cortical networks arises early in development prior to the onset of sensory experience. However, how endogenously generated networks respond to the onset of sensory experience, and how they form mature sensory representations with experience remains unclear. Here we examine this "nature-nurture transform" using in vivo calcium imaging in ferret visual cortex. At eye-opening, visual stimulation evokes robust patterns of cortical activity that are highly variable within and across trials, severely limiting stimulus discriminability. Initial evoked responses are distinct from spontaneous activity of the endogenous network. Visual experience drives the development of low-dimensional, reliable representations aligned with spontaneous activity. A computational model shows that alignment of novel visual inputs and recurrent cortical networks can account for the emergence of reliable visual representations.
In the recent past, we are making huge progress in the field of Artificial Intelligence. Since the rise of neural networks, astonishing new frontiers are continuously being discovered. The development is so fast that overall no major technical limits are in sight. Hence, digitization has expanded from the base of academia and industry to such an extent that it is prevalent in the politics, mass media and even popular arts. The DFG-funded project Specialized Information Service for Biodiversity Research and the BMBF-funded project Linked Open Tafsir can be placed exactly in that overall development. Both projects aim to build an intelligent, up-to-date, modern research infrastructure on biodiversity and theological studies for scholars researching in these respective fields of historical science. Starting from digitized German and Arabic historical literature containing so far unavailable valuable knowledge on biodiversity and theological studies, at its core, our dissertation targets to incorporate state-of-the-art Machine Learning methods for analyzing natural language texts of low-resource languages and enabling foundational Natural Language Processing tasks on them, such as Sentence Boundary Detection, Named Entity Recognition, and Topic Modeling. This ultimately leads to paving the way for new scientific discoveries in the historical disciplines of natural science and humanities. By enriching the landscape of historical low-resource languages with valuable annotation data, our work becomes part of the greater movement of digitizing the society, thus allowing people to focus on things which really matter in science and industry.
Recent experimental findings have reported the presence of unconventional charge orders in the enlarged (2 × 2) unit-cell of kagome metals AV3Sb5 (A = K, Rb, Cs) and hinted towards specific topological signatures. Motivated by these discoveries, we investigate the types of topological phases that can be realized in such kagome superlattices. In this context, we employ a recently introduced statistical method capable of constructing topological models for any generic lattice. By analyzing large data sets generated from symmetry-guided distributions of randomized tight-binding parameters, and labeled with the corresponding topological index, we extract physically meaningful information. We illustrate the possible real-space manifestations of charge and bond modulations and associated flux patterns for different topological classes, and discuss their relation to present theoretical predictions and experimental signatures for the AV3Sb5 family. Simultaneously, we predict higher-order topological phases that may be realized by appropriately manipulating the currently known systems.
TriMem: A parallelized hybrid Monte Carlo software for efficient simulations of lipid membranes
(2022)
Lipid membranes are integral building blocks of living cells and perform a multitude of biological functions. Currently, molecular simulations of cellular-scale membrane remodeling processes at atomic resolution are extremely difficult, due to their size, complexity, and the large times-scales on which these processes occur. Instead, elastic membrane models are used to simulate membrane shapes and transitions between them and to infer their properties and functions. Unfortunately, an efficiently parallelized open-source simulation code to do so has been lacking. Here, we present TriMem, a parallel hybrid Monte Carlo simulation engine for triangulated lipid membranes. The kernels are efficiently coded in C++ and wrapped with Python for ease-of-use. The parallel implementation of the energy and gradient calculations and of Monte Carlo flip moves of edges in the triangulated membrane enable us to simulate large and highly curved membrane structures. For validation, we reproduce phase diagrams of vesicles with varying surface-to-volume ratios and area difference. We also compute the density of states to verify correct Boltzmann sampling. The software can be used to tackle a range of large-scale membrane remodeling processes as a step toward cell-scale simulations. Additionally, extensive documentation make the software accessible to the broad biophysics and computational cell biology communities.
Neural computations emerge from recurrent neural circuits that comprise hundreds to a few thousand neurons. Continuous progress in connectomics, electrophysiology, and calcium imaging require tractable spiking network models that can consistently incorporate new information about the network structure and reproduce the recorded neural activity features. However, it is challenging to predict which spiking network connectivity configurations and neural properties can generate fundamental operational states and specific experimentally reported nonlinear cortical computations. Theoretical descriptions for the computational state of cortical spiking circuits are diverse, including the balanced state where excitatory and inhibitory inputs balance almost perfectly or the inhibition stabilized state (ISN) where the excitatory part of the circuit is unstable. It remains an open question whether these states can co-exist with experimentally reported nonlinear computations and whether they can be recovered in biologically realistic implementations of spiking networks. Here, we show how to identify spiking network connectivity patterns underlying diverse nonlinear computations such as XOR, bistability, inhibitory stabilization, supersaturation, and persistent activity. We established a mapping between the stabilized supralinear network (SSN) and spiking activity which allowed us to pinpoint the location in parameter space where these activity regimes occur. Notably, we found that biologically-sized spiking networks can have irregular asynchronous activity that does not require strong excitation-inhibition balance or large feedforward input and we showed that the dynamic firing rate trajectories in spiking networks can be precisely targeted without error-driven training algorithms.
Mathematical modeling of the molecular switch of TNFR1-mediated signaling pathways using Petri nets
(2021)
The paper describes a mathematical model of the molecular switch of cell survival, apoptosis, and necroptosis in cellular signaling pathways initiated by tumor necrosis factor 1. Based on experimental findings in the current literature, we constructed a Petri net model in terms of detailed molecular reactions for the molecular players, protein complexes, post-translational modifications, and cross talk. The model comprises 118 biochemical entities, 130 reactions, and 299 connecting edges. Applying Petri net analysis techniques, we found 279 pathways describing complete signal flows from receptor activation to cellular response, representing the combinatorial diversity of functional pathways.120 pathways steered the cell to survival, whereas 58 and 35 pathways led to apoptosis and necroptosis, respectively. For 65 pathways, the triggered response was not deterministic, leading to multiple possible outcomes. Based on the Petri net, we investigated the detailed in silico knockout behavior and identified important checkpoints of the TNFR1 signaling pathway in terms of ubiquitination within complex I and the gene expression dependent on NF-κB, which controls the caspase activity in complex II and apoptosis induction.
The human brain achieves visual object recognition through multiple stages of linear and nonlinear transformations operating at a millisecond scale. To predict and explain these rapid transformations, computational neuroscientists employ machine learning modeling techniques. However, state-of-the-art models require massive amounts of data to properly train, and to the present day there is a lack of vast brain datasets which extensively sample the temporal dynamics of visual object recognition. Here we collected a large and rich dataset of high temporal resolution EEG responses to images of objects on a natural background. This dataset includes 10 participants, each with 82,160 trials spanning 16,740 image conditions. Through computational modeling we established the quality of this dataset in five ways. First, we trained linearizing encoding models that successfully synthesized the EEG responses to arbitrary images. Second, we correctly identified the recorded EEG data image conditions in a zero-shot fashion, using EEG synthesized responses to hundreds of thousands of candidate image conditions. Third, we show that both the high number of conditions as well as the trial repetitions of the EEG dataset contribute to the trained models’ prediction accuracy. Fourth, we built encoding models whose predictions well generalize to novel participants. Fifth, we demonstrate full end-to-end training of randomly initialized DNNs that output EEG responses for arbitrary input images. We release this dataset as a tool to foster research in visual neuroscience and computer vision.
The human brain achieves visual object recognition through multiple stages of nonlinear transformations operating at a millisecond scale. To predict and explain these rapid transformations, computational neuroscientists employ machine learning modeling techniques. However, state-of-the-art models require massive amounts of data to properly train, and to the present day there is a lack of vast brain datasets which extensively sample the temporal dynamics of visual object recognition. Here we collected a large and rich dataset of high temporal resolution EEG responses to images of objects on a natural background. This dataset includes 10 participants, each with 82,160 trials spanning 16,740 image conditions. Through computational modeling we established the quality of this dataset in five ways. First, we trained linearizing encoding models that successfully synthesized the EEG responses to arbitrary images. Second, we correctly identified the recorded EEG data image conditions in a zero-shot fashion, using EEG synthesized responses to hundreds of thousands of candidate image conditions. Third, we show that both the high number of conditions as well as the trial repetitions of the EEG dataset contribute to the trained models’ prediction accuracy. Fourth, we built encoding models whose predictions well generalize to novel participants. Fifth, we demonstrate full end-to-end training of randomly initialized DNNs that output M/EEG responses for arbitrary input images. We release this dataset as a tool to foster research in visual neuroscience and computer vision.
Die Emergenz digitaler Netzwerke ist auf die ständige Entwicklung und Transformation neuer Informationstechnologien zurückzuführen.
Dieser Strukturwandel führt zu äußerst komplexen Systemen in vielen verschiedenen Lebensbereichen.
Es besteht daher verstärkt die Notwendigkeit, die zugrunde liegenden wesentlichen Eigenschaften von realen Netzwerken zu untersuchen und zu verstehen.
In diesem Zusammenhang wird die Netzwerkanalyse als Mittel für die Untersuchung von Netzwerken herangezogen und stellt beobachtete Strukturen mithilfe mathematischer Modelle dar.
Hierbei, werden in der Regel parametrisierbare Zufallsgraphen verwendet, um eine systematische experimentelle Evaluation von Algorithmen und Datenstrukturen zu ermöglichen.
Angesichts der zunehmenden Menge an Informationen, sind viele Aspekte der Netzwerkanalyse datengesteuert und zur Interpretation auf effiziente Algorithmen angewiesen.
Algorithmische Lösungen müssen daher sowohl die strukturellen Eigenschaften der Eingabe als auch die Besonderheiten der zugrunde liegenden Maschinen, die sie ausführen, sorgfältig berücksichtigen.
Die Generierung und Analyse massiver Netzwerke ist dementsprechend eine anspruchsvolle Aufgabe für sich.
Die vorliegende Arbeit bietet daher algorithmische Lösungen für die Generierung und Analyse massiver Graphen.
Zu diesem Zweck entwickeln wir Algorithmen für das Generieren von Graphen mit vorgegebenen Knotengraden, die Berechnung von Zusammenhangskomponenten massiver Graphen und zertifizierende Grapherkennung für Instanzen, die die Größe des Hauptspeichers überschreiten.
Unsere Algorithmen und Implementierungen sind praktisch effizient für verschiedene Maschinenmodelle und bieten sequentielle, Shared-Memory parallele und/oder I/O-effiziente Lösungen.
Es geht um Werbung, Betrug oder die Optimierung von Geschäftsmodellen: Verbraucherdaten sind ein kostbares Gut, das Kreditgeber und Versicherer genauso interessiert wie Händler und Kriminelle. Kai Rannenberg, Professor für Mobile Business & Multilateral Security an der Goethe-Universität, forscht zur Cybersicherheit. Dirk Frank hat mit dem Wirtschaftsinformatiker über Datenschutz, Hackerangriffe und das Auto als »Handy auf Rädern« gesprochen.
Das Gebiet der synthetischen Netzwerke boomt. Mithilfe solcher im Computer simulierten Netze werden heute so unterschiedliche Dinge wie die Verknüpfung der Neuronen im Gehirn, der Datenverkehr im Internet oder Stromnetze untersucht. Ein Forscherteam um Ulrich Meyer vom Institut für Informatik der Goethe-Universität hat nun Standardverfahren zur Erstellung solcher Netze einen wichtigen Schritt vorangebracht.
Antimicrobial resistant infections arise as a consequential response to evolutionary mechanisms within microbes which cause them to be protected from the effects of antimicrobials. The frequent occurrence of resistant infections poses a global public health threat as their control has become challenging despite many efforts. The dynamics of such infections are driven by processes at multiple levels. For a long time, mathematical models have proved valuable for unravelling complex mechanisms in the dynamics of infections. In this thesis, we focus on mathematical approaches to modelling the development and spread of resistant infections at between-host (population-wide) and within-host (individual) levels.
Within an individual host, switching between treatments has been identified as one of the methods that can be employed for the gradual eradication of resistant strains on the long term. With this as motivation, we study the problem using dynamical systems and notions from control theory. We present a model based on deterministic logistic differential equations which capture the general dynamics of microbial resistance inside an individual host. Fundamentally, this model describes the spread of resistant infections whilst accounting for evolutionary mutations observed in resistant pathogens and capturing them in mutation matrices. We extend this model to explore the implications of therapy switching from a control theoretic perspective by using switched systems and developing control strategies with the goal of reducing the appearance of drug resistant pathogens within the host.
At the between-host level, we use compartmental models to describe the transmission of infection between multiple individuals in a population. In particular, we make a case study of the evolution and spread of the novel coronavirus (SARS-CoV-2) pandemic. So far, vaccination remains a critical component in the eventual solution to this public health crisis. However, as with many other pathogens, vaccine resistant variants of the virus have been a major concern in control efforts by governments and all stakeholders. Using network theory, we investigate the spread and transmission of the disease on social networks by compartmentalising and studying the progression of the disease in each compartment, considering both the original virus strain and one of its highly transmissible vaccine-resistant mutant strains. We investigate these dynamics in the presence of vaccinations and other interventions. Although vaccinations are of absolute importance during viral outbreaks, resistant variants coupled with population hesitancy towards vaccination can lead to further spread of the virus.
Understanding the role of short-interfering RNA (siRNA) in diverse biological processes is of current interest and often approached through small RNA sequencing. However, analysis of these datasets is difficult due to the complexity of biological RNA processing pathways, which differ between species. Several properties like strand specificity, length distribution, and distribution of soft-clipped bases are few parameters known to guide researchers in understanding the role of siRNAs. We present RAPID, a generic eukaryotic siRNA analysis pipeline, which captures information inherent in the datasets and automatically produces numerous visualizations as user-friendly HTML reports, covering multiple categories required for siRNA analysis. RAPID also facilitates an automated comparison of multiple datasets, with one of the normalization techniques dedicated for siRNA knockdown analysis, and integrates differential expression analysis using DESeq2.
Summary: Understanding the role of short-interfering RNA (siRNA) in diverse biological processes is of current interest and often approached through small RNA sequencing. However, analysis of these datasets is difficult due to the complexity of biological RNA processing pathways, which differ between species. Several properties like strand specificity, length distribution, and distribution of soft-clipped bases are few parameters known to guide researchers in understanding the role of siRNAs. We present RAPID, a generic eukaryotic siRNA analysis pipeline, which captures information inherent in the datasets and automatically produces numerous visualizations as user-friendly HTML reports, covering multiple categories required for siRNA analysis. RAPID also facilitates an automated comparison of multiple datasets, with one of the normalization techniques dedicated for siRNA knockdown analysis, and integrates differential expression analysis using DESeq2. RAPID is available under MIT license at https://github.com/SchulzLab/RAPID. We recommend using it as a conda environment available from https://anaconda.org/bioconda/rapid.
Euclidean distance-optimized data transformation for cluster analysis in biomedical data (EDOtrans)
(2022)
Background: Data transformations are commonly used in bioinformatics data processing in the context of data projection and clustering. The most used Euclidean metric is not scale invariant and therefore occasionally inappropriate for complex, e.g., multimodal distributed variables and may negatively affect the results of cluster analysis. Specifically, the squaring function in the definition of the Euclidean distance as the square root of the sum of squared differences between data points has the consequence that the value 1 implicitly defines a limit for distances within clusters versus distances between (inter-) clusters.
Methods: The Euclidean distances within a standard normal distribution (N(0,1)) follow a N(0,2–√) distribution. The EDO-transformation of a variable X is proposed as EDO=X/(2–√⋅s) following modeling of the standard deviation s by a mixture of Gaussians and selecting the dominant modes via item categorization. The method was compared in artificial and biomedical datasets with clustering of untransformed data, z-transformed data, and the recently proposed pooled variable scaling.
Results: A simulation study and applications to known real data examples showed that the proposed EDO scaling method is generally useful. The clustering results in terms of cluster accuracy, adjusted Rand index and Dunn’s index outperformed the classical alternatives. Finally, the EDO transformation was applied to cluster a high-dimensional genomic dataset consisting of gene expression data for multiple samples of breast cancer tissues, and the proposed approach gave better results than classical methods and was compared with pooled variable scaling.
Conclusions: For multivariate procedures of data analysis, it is proposed to use the EDO transformation as a better alternative to the established z-standardization, especially for nontrivially distributed data. The “EDOtrans” R package is available at https://cran.r-project.org/package=EDOtrans.
Human lymph nodes play a central part of immune defense against infection agents and tumor cells. Lymphoid follicles are compartments of the lymph node which are spherical, mainly filled with B cells. B cells are cellular components of the adaptive immune systems. In the course of a specific immune response, lymphoid follicles pass different morphological differentiation stages. The morphology and the spatial distribution of lymphoid follicles can be sometimes associated to a particular causative agent and development stage of a disease. We report our new approach for the automatic detection of follicular regions in histological whole slide images of tissue sections immuno-stained with actin. The method is divided in two phases: (1) shock filter-based detection of transition points and (2) segmentation of follicular regions. Follicular regions in 10 whole slide images were manually annotated by visual inspection, and sample surveys were conducted by an expert pathologist. The results of our method were validated by comparing with the manual annotation. On average, we could achieve a Zijbendos similarity index of 0.71, with a standard deviation of 0.07.
Consciousness transiently fades away during deep sleep, more stably under anesthesia, and sometimes permanently due to brain injury. The development of an index to quantify the level of consciousness across these different states is regarded as a key problem both in basic and clinical neuroscience. We argue that this problem is ill-defined since such an index would not exhaust all the relevant information about a given state of consciousness. While the level of consciousness can be taken to describe the actual brain state, a complete characterization should also include its potential behavior against external perturbations. We developed and analyzed whole-brain computational models to show that the stability of conscious states provides information complementary to their similarity to conscious wakefulness. Our work leads to a novel methodological framework to sort out different brain states by their stability and reversibility, and illustrates its usefulness to dissociate between physiological (sleep), pathological (brain-injured patients), and pharmacologically-induced (anesthesia) loss of consciousness.
Tracking influenza a virus infection in the lung from hematological data with machine learning
(2022)
The tracking of pathogen burden and host responses with minimal-invasive methods during respiratory infections is central for monitoring disease development and guiding treatment decisions. Utilizing a standardized murine model of respiratory Influenza A virus (IAV) infection, we developed and tested different supervised machine learning models to predict viral burden and immune response markers, i.e. cytokines and leukocytes in the lung, from hematological data. We performed independently in vivo infection experiments to acquire extensive data for training and testing purposes of the models. We show here that lung viral load, neutrophil counts, cytokines like IFN-γ and IL-6, and other lung infection markers can be predicted from hematological data. Furthermore, feature analysis of the models shows that blood granulocytes and platelets play a crucial role in prediction and are highly involved in the immune response against IAV. The proposed in silico tools pave the path towards improved tracking and monitoring of influenza infections and possibly other respiratory infections based on minimal-invasively obtained hematological parameters.
Changes in the efficacies of synapses are thought to be the neurobiological basis of learning and memory. The efficacy of a synapse depends on its current number of neurotransmitter receptors. Recent experiments have shown that these receptors are highly dynamic, moving back and forth between synapses on time scales of seconds and minutes. This suggests spontaneous fluctuations in synaptic efficacies and a competition of nearby synapses for available receptors. Here we propose a mathematical model of this competition of synapses for neurotransmitter receptors from a local dendritic pool. Using minimal assumptions, the model produces a fast multiplicative scaling behavior of synapses. Furthermore, the model explains a transient form of heterosynaptic plasticity and predicts that its amount is inversely related to the size of the local receptor pool. Overall, our model reveals logistical tradeoffs during the induction of synaptic plasticity due to the rapid exchange of neurotransmitter receptors between synapses.
Changes in the efficacies of synapses are thought to be the neurobiological basis of learning and memory. The efficacy of a synapse depends on its current number of neurotransmitter receptors. Recent experiments have shown that these receptors are highly dynamic, moving back and forth between synapses on time scales of seconds and minutes. This suggests spontaneous fluctuations in synaptic efficacies and a competition of nearby synapses for available receptors. Here we propose a mathematical model of this competition of synapses for neurotransmitter receptors from a local dendritic pool. Using minimal assumptions, the model produces a fast multiplicative scaling behavior of synapses. Furthermore, the model explains a transient form of heterosynaptic plasticity and predicts that its amount is inversely related to the size of the local receptor pool. Overall, our model reveals logistical tradeoffs during the induction of synaptic plasticity due to the rapid exchange of neurotransmitter receptors between synapses.
Bacteria of the genera Photorhabdus and Xenorhabdus produce a plethora of natural products to support their similar symbiotic lifecycles. For many of these compounds, the specific bioactivities are unknown. One common challenge in natural product research when trying to prioritize research efforts is the rediscovery of identical (or highly similar) compounds from different strains. Linking genome sequence to metabolite production can help in overcoming this problem. However, sequences are typically not available for entire collections of organisms. Here we perform a comprehensive metabolic screening using HPLC-MS data associated with a 114-strain collection (58 Photorhabdus and 56 Xenorhabdus) from across Thailand and explore the metabolic variation among the strains, matched with several abiotic factors. We utilize machine learning in order to rank the importance of individual metabolites in determining all given metadata. With this approach, we were able to prioritize metabolites in the context of natural product investigations, leading to the identification of previously unknown compounds. The top three highest-ranking features were associated with Xenorhabdus and attributed to the same chemical entity, cyclo(tetrahydroxybutyrate). This work addresses the need for prioritization in high-throughput metabolomic studies and demonstrates the viability of such an approach in future research.
People can describe spatial scenes with language and, vice versa, create images based on linguistic descriptions. However, current systems do not even come close to matching the complexity of humans when it comes to reconstructing a scene from a given text. Even the ever-advancing development of better and better Transformer-based models has not been able to achieve this so far. This task, the automatic generation of a 3D scene based on an input text, is called text-to-3D scene generation. The key challenge, and focus of this dissertation, now relate to the following topics:
(a) Analyses of how well current language models understand spatial information, how static embeddings compare, and whether they can be improved by anaphora resolution.
(b) Automated resource generation for context expansion and grounding that can help in the creation of realistic scenes.
(c) Creation of a VR-based text-to-3D scene system that can be used as an annotation and active-learning environment, but can also be easily extended in a modular way with additional features to solve more contexts in the future.
(d) Analyze existing practices and tools for digital and virtual teaching, learning, and collaboration, as well as the conditions and strategies in the context of VR.
In the first part of this work, we could show that static word embeddings do not benefit significantly from pronoun substitution. We explain this result by the loss of contextual information, the reduction in the relative occurrence of rare words, and the absence of pronouns to be substituted. But we were able to we have shown that both static and contextualizing language models appear to encode object knowledge, but require a sophisticated apparatus to retrieve it. The models themselves in combination with the measures differ greatly in terms of the amount of knowledge they allow to extract.
Classifier-based variants perform significantly better than the unsupervised methods from bias research, but this is also due to overfitting. The resources generated for this evaluation are later also an important component of point three.
In the second part, we present AffordanceUPT, a modularization of UPT trained on the HICO-DET dataset, which we have extended with Gibsonien/telic annotations. We then show that AffordanceUPT can effectively make the Gibsonian/telic distinction and that the model learns other correlations in the data to make such distinctions (e.g., the presence of hands in the image) that have important implications for grounding images to language.
The third part first presents a VR project to support spatial annotation respectively IsoSpace. The direct spatial visualization and the immediate interaction with the 3D objects should make the labeling more intuitive and thus easier. The project will later be incorporated as part of the Semantic Scene Builder (SeSB). The project itself in turn relies on the Text2SceneVR presented here for generating spatial hypertext, which in turn is based on the VAnnotatoR. Finally, we introduce Semantic Scene Builder (SeSB), a VR-based text-to-3D scene framework using Semantic Annotation Framework (SemAF) as a scheme for annotating semantic relations. It integrates a wide range of tools and resources by utilizing SemAF and UIMA as a unified data structure to generate 3D scenes from textual descriptions and also supports annotations. When evaluating SeSB against another state-of-the-art tool, it was found that our approach not only performed better, but also allowed us to model a wider variety of scenes. The final part reviews existing practices and tools for digital and virtual teaching, learning, and collaboration, as well as the conditions and strategies needed to make the most of technological opportunities in the future.
Summary We introduce fsbrain, an R package for the visualization of neuroimaging data. The package can be used to visualize vertex-wise and region-wise morphometry data, parcellations, labels and statistical results on brain surfaces in three dimensions (3D). Voxel data can be displayed in lightbox mode. The fsbrain package offers various customization options and produces publication quality plots which can be displayed interactively, saved as bitmap images, or integrated into R notebooks.
Availability and Implementation The software, source code and documentation are available under the MIT license at https://github.com/dfsp-spirit/fsbrain. Releases can be installed directly from the Comprehensive R Archive Network (CRAN).
Grasping the meaning of everyday visual events is a fundamental feat of human intelligence that hinges on diverse neural processes ranging from vision to higher-level cognition. Deciphering the neural basis of visual event understanding requires rich, extensive, and appropriately designed experimental data. However, this type of data is hitherto missing. To fill this gap, we introduce the BOLD Moments Dataset (BMD), a large dataset of whole-brain fMRI responses to over 1,000 short (3s) naturalistic video clips and accompanying metadata. We show visual events interface with an array of processes, extending even to memory, and we reveal a match in hierarchical processing between brains and video-computable deep neural networks. Furthermore, we showcase that BMD successfully captures temporal dynamics of visual events at second resolution. BMD thus establishes a critical groundwork for investigations of the neural basis of visual event understanding.
Visual scene perception is mediated by a set of cortical regions that respond preferentially to images of scenes, including the occipital place area (OPA) and parahippocampal place area (PPA). However, the differential contribution of OPA and PPA to scene perception remains an open research question. In this study, we take a deep neural network (DNN)-based computational approach to investigate the differences in OPA and PPA function. In a first step we search for a computational model that predicts fMRI responses to scenes in OPA and PPA well. We find that DNNs trained to predict scene components (e.g., wall, ceiling, floor) explain higher variance uniquely in OPA and PPA than a DNN trained to predict scene category (e.g., bathroom, kitchen, office). This result is robust across several DNN architectures. On this basis, we then determine whether particular scene components predicted by DNNs differentially account for unique variance in OPA and PPA. We find that variance in OPA responses uniquely explained by the navigation-related floor component is higher compared to the variance explained by the wall and ceiling components. In contrast, PPA responses are better explained by the combination of wall and floor, that is scene components that together contain the structure and texture of the scene. This differential sensitivity to scene components suggests differential functions of OPA and PPA in scene processing. Moreover, our results further highlight the potential of the proposed computational approach as a general tool in the investigation of the neural basis of human scene perception.
The human visual cortex enables visual perception through a cascade of hierarchical computations in cortical regions with distinct functionalities. Here, we introduce an AI-driven approach to discover the functional mapping of the visual cortex. We related human brain responses to scene images measured with functional MRI (fMRI) systematically to a diverse set of deep neural networks (DNNs) optimized to perform different scene perception tasks. We found a structured mapping between DNN tasks and brain regions along the ventral and dorsal visual streams. Low-level visual tasks mapped onto early brain regions, 3-dimensional scene perception tasks mapped onto the dorsal stream, and semantic tasks mapped onto the ventral stream. This mapping was of high fidelity, with more than 60% of the explainable variance in nine key regions being explained. Together, our results provide a novel functional mapping of the human visual cortex and demonstrate the power of the computational approach.
Gasdermin-D (GSDMD) is the ultimate effector of pyroptosis, a form of programmed cell death associated with pathogen invasion and inflammation. After proteolytic cleavage by caspases activated by the inflammasome, the GSDMD N-terminal domain (GSDMDNT) assembles on the inner leaflet of the plasma membrane and induces the formation of large membrane pores. We use atomistic molecular dynamics simulations to study GSDMDNT monomers, oligomers, and rings in an asymmetric plasma membrane mimetic. We identify distinct interaction motifs of GSDMDNT with phosphatidylinositol-4,5-bisphosphate (PI(4,5)P2) and phosphatidylserine (PS) head-groups and describe differential lipid binding between the pore and prepore conformations. Oligomers are stabilized by shared lipid binding sites between neighboring monomers acting akin to double-sided tape. We show that already small GSDMDNT oligomers form stable, water-filled and ion-conducting membrane pores bounded by curled beta-sheets. In large-scale simulations, we resolve the process of pore formation by lipid detachment from GSDMDNT arcs and lipid efflux from partial rings. We find that that high-order GSDMDNT oligomers can crack under the line tension of 86 pN created by an open membrane edge to form the slit pores or closed GSDMDNT rings seen in experiment. Our simulations provide a detailed view of key steps in GSDMDNT-induced plasma membrane pore formation, including sublytic pores that explain nonselective ion flux during early pyroptosis.
Gasdermin-D (GSDMD) is the ultimate effector of pyroptosis, a form of programmed cell death associated with pathogen invasion and inflammation. After proteolytic cleavage by caspases, the GSDMD N-terminal domain (GSDMDNT) assembles on the inner leaflet of the plasma membrane and induces the formation of membrane pores. We use atomistic molecular dynamics simulations to study GSDMDNT monomers, oligomers, and rings in an asymmetric plasma membrane mimetic. We identify distinct interaction motifs of GSDMDNT with phosphatidylinositol-4,5-bisphosphate (PI(4,5)P2) and phosphatidylserine (PS) headgroups and describe their conformational dependence. Oligomers are stabilized by shared lipid binding sites between neighboring monomers acting akin to double-sided tape. We show that already small GSDMDNT oligomers support stable, water-filled, and ion-conducting membrane pores bounded by curled beta-sheets. In large-scale simulations, we resolve the process of pore formation from GSDMDNT arcs and lipid efflux from partial rings. We find that high-order GSDMDNT oligomers can crack under the line tension of 86 pN created by an open membrane edge to form the slit pores or closed GSDMDNT rings seen in atomic force microscopy experiments. Our simulations provide a detailed view of key steps in GSDMDNT-induced plasma membrane pore formation, including sublytic pores that explain nonselective ion flux during early pyroptosis.
Nuclear pore complexes (NPCs) mediate nucleocytoplasmic transport. Their intricate 120 MDa architecture remains incompletely understood. Here, we report a near-complete structural model of the human NPC scaffold with explicit membrane and in multiple conformational states. We combined AI-based structure prediction with in situ and in cellulo cryo-electron tomography and integrative modeling. We show that linker Nups spatially organize the scaffold within and across subcomplexes to establish the higher-order structure. Microsecond-long molecular dynamics simulations suggest that the scaffold is not required to stabilize the inner and outer nuclear membrane fusion, but rather widens the central pore. Our work exemplifies how AI-based modeling can be integrated with in situ structural biology to understand subcellular architecture across spatial organization levels.
Inspired by the physiology of neuronal systems in the brain, artificial neural networks have become an invaluable tool for machine learning applications. However, their biological realism and theoretical tractability are limited, resulting in poorly understood parameters. We have recently shown that biological neuronal firing rates in response to distributed inputs are largely independent of size, meaning that neurons are typically responsive to the proportion, not the absolute number, of their inputs that are active. Here we introduce such a normalisation, where the strength of a neuron’s afferents is divided by their number, to various sparsely-connected artificial networks. The learning performance is dramatically increased, providing an improvement over other widely-used normalisations in sparse networks. The resulting machine learning tools are universally applicable and biologically inspired, rendering them better understood and more stable in our tests.
Orientation hypercolumns in the visual cortex are delimited by the repeating pinwheel patterns of orientation selective neurons. We design a generative model for visual cortex maps that reproduces such orientation hypercolumns as well as ocular dominance maps while preserving retinotopy. The model uses a neural placement method based on t–distributed stochastic neighbour embedding (t–SNE) to create maps that order common features in the connectivity matrix of the circuit. We find that, in our model, hypercolumns generally appear with fixed cell numbers independently of the overall network size. These results would suggest that existing differences in absolute pinwheel densities are a consequence of variations in neuronal density. Indeed, available measurements in the visual cortex indicate that pinwheels consist of a constant number of ∼30, 000 neurons. Our model is able to reproduce a large number of characteristic properties known for visual cortex maps. We provide the corresponding software in our MAPStoolbox for Matlab.
Artificial neural networks, taking inspiration from biological neurons, have become an invaluable tool for machine learning applications. Recent studies have developed techniques to effectively tune the connectivity of sparsely-connected artificial neural networks, which have the potential to be more computationally efficient than their fully-connected counterparts and more closely resemble the architectures of biological systems. We here present a normalisation, based on the biophysical behaviour of neuronal dendrites receiving distributed synaptic inputs, that divides the weight of an artificial neuron’s afferent contacts by their number. We apply this dendritic normalisation to various sparsely-connected feedforward network architectures, as well as simple recurrent and self-organised networks with spatially extended units. The learning performance is significantly increased, providing an improvement over other widely-used normalisations in sparse networks. The results are two-fold, being both a practical advance in machine learning and an insight into how the structure of neuronal dendritic arbours may contribute to computation.
The electrical and computational properties of neurons in our brains are determined by a rich repertoire of membrane-spanning ion channels and elaborate dendritic trees. However, the precise reason for this inherent complexity remains unknown. Here, we generated large stochastic populations of biophysically realistic hippocampal granule cell models comparing those with all 15 ion channels to their reduced but functional counterparts containing only 5 ion channels. Strikingly, valid parameter combinations in the full models were more frequent and more stable in the face of perturbations to channel expression levels. Scaling up the numbers of ion channels artificially in the reduced models recovered these advantages confirming the key contribution of the actual number of ion channel types. We conclude that the diversity of ion channels gives a neuron greater flexibility and robustness to achieve target excitability.
The measurement of protein dynamics by proteomics to study cell remodeling has seen increased attention over the last years. This development is largely driven by a number of technological advances in proteomics methods. Pulsed stable isotope labeling in cell culture (SILAC) combined with tandem mass tag (TMT) labeling has evolved as a gold standard for profiling protein synthesis and degradation. While the experimental setup is similar to typical proteomics experiments, the data analysis proves more difficult: After peptide identification through search engines, data extraction requires either custom scripted pipelines or tedious manual table manipulations to extract the TMT-labeled heavy and light peaks of interest. To overcome this limitation, which deters researchers from using protein dynamic proteomics, we developed a user-friendly, browser-based application that allows easy and reproducible data analysis without the need for scripting experience. In addition, we provide a python package that can be implemented in established data analysis pipelines. We anticipate that this tool will ease data analysis and spark further research aimed at monitoring protein translation and degradation by proteomics.
Diminished sense of smell impairs the quality of life but olfactorily disabled people are hardly considered in measures of disability inclusion. We aimed to stratify perceptual characteristics and odors according to the extent to which they are perceived differently with reduced sense of smell, as a possible basis for creating olfactory experiences that are enjoyed in a similar way by subjects with normal or impaired olfactory function. In 146 subjects with normal or reduced olfactory function, perceptual characteristics (edibility, intensity, irritation, temperature, familiarity, hedonics, painfulness) were tested for four sets of 10 different odors each. Data were analyzed with (i) a projection based on principal component analysis and (ii) the training of a machine-learning algorithm in a 1000-fold cross-validated setting to distinguish between olfactory diagnosis based on odor property ratings. Both analytical approaches identified perceived intensity and familiarity with the odor as discriminating characteristics between olfactory diagnoses, while evoked pain sensation and perceived temperature were not discriminating, followed by edibility. Two disjoint sets of odors were identified, i.e., d = 4 “discriminating odors” with respect to olfactory diagnosis, including cis-3-hexenol, methyl salicylate, 1-butanol and cineole, and d = 7 “non-discriminating odors”, including benzyl acetate, heptanal, 4-ethyl-octanoic acid, methional, isobutyric acid, 4-decanolide and p-cresol. Different weightings of the perceptual properties of odors with normal or reduced sense of smell indicate possibilities to create sensory experiences such as food, meals or scents that by emphasizing trigeminal perceptions can be enjoyed by both normosmic and hyposmic individuals.
In the human brain, the incoming light to the retina is transformed into meaningful representations that allow us to interact with the world. In a similar vein, the RGB pixel values are transformed by a deep neural network (DNN) into meaningful representations relevant to solving a computer vision task it was trained for. Therefore, in my research, I aim to reveal insights into the visual representations in the human visual cortex and DNNs solving vision tasks.
In the previous decade, DNNs have emerged as the state-of-the-art models for predicting neural responses in the human and monkey visual cortex. Research has shown that training on a task related to a brain region’s function leads to better predictivity than a randomly initialized network. Based on this observation, we proposed that we can use DNNs trained on different computer vision tasks to identify functional mapping of the human visual cortex.
To validate our proposed idea, we first investigate a brain region occipital place area (OPA) using DNNs trained on scene parsing task and scene classification task. From the previous investigations about OPA’s functions, we knew that it encodes navigational affordances that require spatial information about the scene. Therefore, we hypothesized that OPA’s representation should be closer to a scene parsing model than a scene classification model as the scene parsing task explicitly requires spatial information about the scene. Our results showed that scene parsing models had representation closer to OPA than scene classification models thus validating our approach.
We then selected multiple DNNs performing a wide range of computer vision tasks ranging from low-level tasks such as edge detection, 3D tasks such as surface normals, and semantic tasks such as semantic segmentation. We compared the representations of these DNNs with all the regions in the visual cortex, thus revealing the functional representations of different regions of the visual cortex. Our results highly converged with previous investigations of these brain regions validating the feasibility of the proposed approach in finding functional representations of the human brain. Our results also provided new insights into underinvestigated brain regions that can serve as starting hypotheses and promote further investigation into those brain regions.
We applied the same approach to find representational insights about the DNNs. A DNN usually consists of multiple layers with each layer performing a computation leading to the final layer that performs prediction for a given task. Training on different tasks could lead to very different representations. Therefore, we first investigate at which stage does the representation in DNNs trained on different tasks starts to differ. We further investigate if the DNNs trained on similar tasks lead to similar representations and on dissimilar tasks lead to more dissimilar representations. We selected the same set of DNNs used in the previous work that were trained on the Taskonomy dataset on a diverse range of 2D, 3D and semantic tasks. Then, given a DNN trained on a particular task, we compared the representation of multiple layers to corresponding layers in other DNNs. From this analysis, we aimed to reveal where in the network architecture task-specific representation is prominent. We found that task specificity increases as we go deeper into the DNN architecture and similar tasks start to cluster in groups. We found that the grouping we found using representational similarity was highly correlated with grouping based on transfer learning thus creating an interesting application of the approach to model selection in transfer learning.
During previous works, several new measures were introduced to compare DNN representations. So, we identified the commonalities in different measures and unified different measures into a single framework referred to as duality diagram similarity. This work opens up new possibilities for similarity measures to understand DNN representations. While demonstrating a much higher correlation with transfer learning than previous state-of-the-art measures we extend it to understanding layer-wise representations of models trained on the Imagenet and Places dataset using different tasks and demonstrate its applicability to layer selection for transfer learning.
In all the previous works, we used the task-specific DNN representations to understand the representations in the human visual cortex and other DNNs. We were able to interpret our findings in terms of computer vision tasks such as edge detection, semantic segmentation, depth estimation, etc. however we were not able to map the representations to human interpretable concepts. Therefore in our most recent work, we developed a new method that associates individual artificial neurons with human interpretable concepts.
Overall, the works in this thesis revealed new insights into the representation of the visual cortex and DNNs...
Dendritic spines are considered a morphological proxy for excitatory synapses, rendering them a target of many different lines of research. Over recent years, it has become possible to image simultaneously large numbers of dendritic spines in 3D volumes of neural tissue. In contrast, currently no automated method for spine detection exists that comes close to the detection performance reached by human experts. However, exploiting such datasets requires new tools for the fully automated detection and analysis of large numbers of spines. Here, we developed an efficient analysis pipeline to detect large numbers of dendritic spines in volumetric fluorescence imaging data. The core of our pipeline is a deep convolutional neural network, which was pretrained on a general-purpose image library, and then optimized on the spine detection task. This transfer learning approach is data efficient while achieving a high detection precision. To train and validate the model we generated a labelled dataset using five human expert annotators to account for the variability in human spine detection. The pipeline enables fully automated dendritic spine detection and reaches a near human-level detection performance. Our method for spine detection is fast, accurate and robust, and thus well suited for large-scale datasets with thousands of spines. The code is easily applicable to new datasets, achieving high detection performance, even without any retraining or adjustment of model parameters.
Active efficient coding explains the development of binocular vision and its failure in amblyopia
(2020)
The development of vision during the first months of life is an active process that comprises the learning of appropriate neural representations and the learning of accurate eye movements. While it has long been suspected that the two learning processes are coupled, there is still no widely accepted theoretical framework describing this joint development. Here, we propose a computational model of the development of active binocular vision to fill this gap. The model is based on a formulation of the active efficient coding theory, which proposes that eye movements as well as stimulus encoding are jointly adapted to maximize the overall coding efficiency. Under healthy conditions, the model self-calibrates to perform accurate vergence and accommodation eye movements. It exploits disparity cues to deduce the direction of defocus, which leads to coordinated vergence and accommodation responses. In a simulated anisometropic case, where the refraction power of the two eyes differs, an amblyopia-like state develops in which the foveal region of one eye is suppressed due to inputs from the other eye. After correcting for refractive errors, the model can only reach healthy performance levels if receptive fields are still plastic, in line with findings on a critical period for binocular vision development. Overall, our model offers a unifying conceptual framework for understanding the development of binocular vision.
Treatments for amblyopia focus on vision therapy and patching of one eye. Predicting the success of these methods remains difficult, however. Recent research has used binocular rivalry to monitor visual cortical plasticity during occlusion therapy, leading to a successful prediction of the recovery rate of the amblyopic eye. The underlying mechanisms and their relation to neural homeostatic plasticity are not known. Here we propose a spiking neural network to explain the effect of short-term monocular deprivation on binocular rivalry. The model reproduces perceptual switches as observed experimentally. When one eye is occluded, inhibitory plasticity changes the balance between the eyes and leads to longer dominance periods for the eye that has been deprived. The model suggests that homeostatic inhibitory plasticity is a critical component of the observed effects and might play an important role in the recovery from amblyopia.
Motivation DNA CpG methylation (CpGm) has proven to be a crucial epigenetic factor in the gene regulatory system. Assessment of DNA CpG methylation values via whole-genome bisulfite sequencing (WGBS) is, however, computationally extremely demanding.
Results We present FAst MEthylation calling (FAME), the first approach to quantify CpGm values directly from bulk or single-cell WGBS reads without intermediate output files. FAME is very fast but as accurate as standard methods, which first produce BS alignment files before computing CpGm values. We present experiments on bulk and single-cell bisulfite datasets in which we show that data analysis can be significantly sped-up and help addressing the current WGBS analysis bottleneck for large-scale datasets without compromising accuracy.
Availability An implementation of FAME is open source and licensed under GPL-3.0 at https://github.com/FischerJo/FAME.
Sample-based longitudinal discrete choice experiments: preferences for electric vehicles over time
(2021)
Discrete choice experiments have emerged as the state-of-the-art method for measuring preferences, but they are mostly used in cross-sectional studies. In seeking to make them applicable for longitudinal studies, our study addresses two common challenges: working with different respondents and handling altering attributes. We propose a sample-based longitudinal discrete choice experiment in combination with a covariate-extended hierarchical Bayes logit estimator that allows one to test the statistical significance of changes. We showcase this method’s use in studies about preferences for electric vehicles over six years and empirically observe that preferences develop in an unpredictable, non-monotonous way. We also find that inspecting only the absolute differences in preferences between samples may result in misleading inferences. Moreover, surveying a new sample produced similar results as asking the same sample of respondents over time. Finally, we experimentally test how adding or removing an attribute affects preferences for the other attributes.
Der Sammelband "Digitale Barrierefreiheit in der Bildung weiter denken: Innovative Impulse aus Praxis, Technik und Didaktik", der von #DigiBar, dem "Netzwerk digitale Barrierefreiheit an hessischen Hochschulen", herausgegeben wird, geht der Frage nach, welchen Status Quo, welche Herausforderungen sowie praktischen und theoretischen Lösungsansätze das Thema digitale Barrierefreiheit in der Bildung aufzeigt. Die Beiträge des vorliegenden Sammelbandes tragen diesem Forschungsdesiderat Rechnung. Die von "studiumdigitale" (Goethe-Universität Frankfurt) in Kooperation mit BliZ, dem Zentrum für blinde und sehbehinderte Studierende (THM - Technischen Hochschule Mittelhessen) herausgegebene Veröffentlichung richtet sich an Lehrende, Hochschulangehörige, Entscheidungsträger*innen, Webadmins, Tutor*innen, soziale Träger*innen und Interessierte. Die Beiträge beleuchten das Titelthema in all seinen Facetten: Von Modellprojekten über Fallstudien und technische Lösungsszenarien bis hin zur Vermittlung praktischer Informationen und zu Szenarien der Barrierefreiheit im multimedialen Raum sind hier Grundlagenforschung, Best-Practice-Beispiele und technische Lösungen in einem Band vereint. Selbstverständlich ist die digitale Veröffentlichung in barrierefreier Form, um ein Vorbild für weitere derartige Veröffentlichungen zu sein. Helfen auch Sie mit, Barrieren in der Bildungswelt abzubauen und Teilhabe zu ermöglichen.
AI-based computer vision systems play a crucial role in the environment perception for autonomous driving. Although the development of self-driving systems has been pursued for multiple decades, it is only recently that breakthroughs in Deep Neural Networks (DNNs) have led to their widespread application in perception pipelines, which are getting more and more sophisticated. However, with this rising trend comes the need for a systematic safety analysis to evaluate the DNN's behavior in difficult scenarios as well as to identify the various factors that cause misbehavior in such systems. This work aims to deliver a crucial contribution to the lacking literature on the systematic analysis of Performance Limiting Factors (PLFs) for DNNs by investigating the task of pedestrian detection in urban traffic from a monocular camera mounted on an autonomous vehicle. To investigate the common factors that lead to DNN misbehavior, six commonly used state-of-the-art object detection architectures and three detection tasks are studied using a new large-scale synthetic dataset and a smaller real-world dataset for pedestrian detection. The systematic analysis includes 17 factors from the literature and four novel factors that are introduced as part of this work. Each of the 21 factors is assessed based on its influence on the detection performance and whether it can be considered a Performance Limiting Factor (PLF). In order to support the evaluation of the detection performance, a novel and task-oriented Pedestrian Detection Safety Metric (PDSM) is introduced, which is specifically designed to aid in the identification of individual factors that contribute to DNN failure. This work further introduces a training approach for F1-Score maximization whose purpose is to ensure that the DNNs are assessed at their highest performance. Moreover, a new occlusion estimation model is introduced to replace the missing pedestrian occlusion annotations in the real-world dataset. Based on a qualitative analysis of the correlation graphs that visualize the correlation between the PLFs and the detection performance, this study identified 16 of the initial 21 factors as being PLFs for DNNs out of which the entropy, the occlusion ratio, the boundary edge strength, and the bounding box aspect ratio turned out to be most severely affecting the detection performance. The findings of this study highlight some of the most serious shortcomings of current DNNs and pave the way for future research to address these issues.
ChatGPT, der Prototyp eines Chatbot, von dem amerikanischen Unternehmen OpenAI entwickelt, ist im Augenblick in aller Munde. Gefragt wird auch: Stellt diese Software eine Herausforderung für den Bildungsbereich dar, werden künftig damit Haus- und Abschlussarbeiten erstellt? Prof. Uwe Walz, Professor für VWL, insbesondere Industrieökonomie an der Goethe-Universität, hat den Chatbot bereits im laufenden Wintersemester mit Studierenden analysiert.
Non-Fungible Token und die Blockchain Technologie haben in dem vergangenen Jahr immer mehr an Popularität gewonnen. Wie bei jeder neuartigen Technologie stellt sich jedoch die Frage, in welchen Bereichen diese eine Anwendung finden können.
Das Ziel in der vorliegenden Arbeit ist es zu beantworten, ob Non-Fungible Token und die Blockchain Technologie eine sinnvolle Anwendung im Bereich von akademischen Zertifikaten hat.
Um diese Frage zu beantworten, sind Gründe für die Anwendung von Non-Fungible Token gegen Nachteile abgewogen und Lösungsansätze für potentielle Risiken erhoben worden. Außerdem wurde selbstständig ein ERC-721 Token Contract für akademische Zertifikate mittels Solidity entwickelt.
Die Arbeit zeigt, dass Blockchain basierte akademische Zertifikate vor allem die Mobilität von Studenten unterstützen, den administrativen Aufwand der Ausstellung und Verifizierung von Abschlusszeugnissen verringern und entgegen der Fälschung von Abschlüssen arbeiten. Außerdem können erwägte Risiken und Nachteile durch Zusammenschluss von Institutionen zu einer Konsortialen Blockchain umgangen werden.
Die erfolgreiche Entwicklung des ERC-721 Token Contracts “MetaDip” zeigt eine potentielle Umsetzung für die Digitalisierung von Abschlusszeugnissen und demonstriert, dass Non-Fungible Token basierte akademische Zertifikate aktuell bereits technisch realisierbar sind.
Die Arbeit legt dar, dass Non-Fungible Token und die Blockchain Technologie eine vielversprechende Zukunft für akademische Zertifikate bietet und bereits von vereinzelten Institutionen realisiert wird. Jedoch müssen noch einige Vorkehrungen getroffen werden, bevor eine breite Umsetzung von Blockchain basierten akademischen Zertifikaten möglich ist.
In this paper, we introduce an approach for future frames prediction based on a single input image. Our method is able to generate an entire video sequence based on the information contained in the input frame. We adopt an autoregressive approach in our generation process, i.e., the output from each time step is fed as the input to the next step. Unlike other video prediction methods that use “one shot” generation, our method is able to preserve much more details from the input image, while also capturing the critical pixel-level changes between the frames. We overcome the problem of generation quality degradation by introducing a “complementary mask” module in our architecture, and we show that this allows the model to only focus on the generation of the pixels that need to be changed, and to reuse those that should remain static from its previous frame. We empirically validate our methods against various video prediction models on the UT Dallas Dataset, and show that our approach is able to generate high quality realistic video sequences from one static input image. In addition, we also validate the robustness of our method by testing a pre-trained model on the unseen ADFES facial expression dataset. We also provide qualitative results of our model tested on a human action dataset: The Weizmann Action database.
Background: An essential step in any medical research project after identifying the research question is to determine if there are sufficient patients available for a study and where to find them. Pursuing digital feasibility queries on available patient data registries has proven to be an excellent way of reusing existing real-world data sources. To support multicentric research, these feasibility queries should be designed and implemented to run across multiple sites and securely access local data. Working across hospitals usually involves working with different data formats and vocabularies. Recently, the Fast Healthcare Interoperability Resources (FHIR) standard was developed by Health Level Seven to address this concern and describe patient data in a standardized format. The Medical Informatics Initiative in Germany has committed to this standard and created data integration centers, which convert existing data into the FHIR format at each hospital. This partially solves the interoperability problem; however, a distributed feasibility query platform for the FHIR standard is still missing.
Objective: This study described the design and implementation of the components involved in creating a cross-hospital feasibility query platform for researchers based on FHIR resources. This effort was part of a large COVID-19 data exchange platform and was designed to be scalable for a broad range of patient data.
Methods: We analyzed and designed the abstract components necessary for a distributed feasibility query. This included a user interface for creating the query, backend with an ontology and terminology service, middleware for query distribution, and FHIR feasibility query execution service.
Results: We implemented the components described in the Methods section. The resulting solution was distributed to 33 German university hospitals. The functionality of the comprehensive network infrastructure was demonstrated using a test data set based on the German Corona Consensus Data Set. A performance test using specifically created synthetic data revealed the applicability of our solution to data sets containing millions of FHIR resources. The solution can be easily deployed across hospitals and supports feasibility queries, combining multiple inclusion and exclusion criteria using standard Health Level Seven query languages such as Clinical Quality Language and FHIR Search. Developing a platform based on multiple microservices allowed us to create an extendable platform and support multiple Health Level Seven query languages and middleware components to allow integration with future directions of the Medical Informatics Initiative.
Conclusions: We designed and implemented a feasibility platform for distributed feasibility queries, which works directly on FHIR-formatted data and distributed it across 33 university hospitals in Germany. We showed that developing a feasibility platform directly on the FHIR standard is feasible.
The main task of modern large experiments with heavy ions, such as CBM (FAIR), STAR (BNL) and ALICE (CERN) is a detailed study of the phase diagram of quantum chromodynamics (QCD) in the quark-gluon plasma (QGP), the equation of state of matter at extremely high baryonic densities, and the transition from the hadronic phase of matter to the quark-gluon phase.
In the thesis, the missing mass method is developed for the reconstruction of short-lived particles with neutral particles in their decay products, as well as its implementation in the form of fast algorithms and a set of software for prac- tical application in heavy ion physics experiments. Mathematical procedures implementing the method were developed and implemented within the KF Par- ticle Finder package for the future CBM (FAIR) experiment and subsequently adapted and applied for processing and analysis of real data in the STAR (BNL) experiment.
The KF Particle Finder package is designed to reconstruct most signal particles from the physics program of the CBM experiment, including strange particles, strange resonances, hypernuclei, light vector mesons, charm particles and char- monium. The package includes searches for over a hundred decays of short-lived particles. This makes the KF Particle Finder a universal platform for short-lived particle reconstruction and physics analysis both online and offline.
The missing mass method has been proposed to reconstruct decays of short-lived charged particles when one of the daughter particles is neutral and is not regis- tered in the detector system. The implementation of the missing mass method was integrated into the KF Particle Finder package to search for 18 decays with a neutral daughter particle.
Like all other algorithms of the KF Particle Finder package, the missing mass method is implemented with extensive use of vector (SIMD) instructions and is optimized for parallel operation on modern many-core high performance com- puter clusters, which can include both processors and coprocessors. A set of algorithms implementing the method was tested on computers with tens of cores and showed high speed and practically linear scalability with respect to the num- ber of cores involved.
It is extremely important, especially for the initial stage of the CBM experiment, which is planned for 2025, to demonstrate already now on real data the reliability of the developed approach, as well as the high efficiency of the current implemen- tation of both the entire KF Particle Finder package, and its integral part, the missing mass method. Such an opportunity was provided by the FAIR Phase-0 program, motivating the use in the STAR experiment of software packages orig- inally developed for the CBM experiment.
Application of the method to real data of the STAR experiment shows very good results with a high signal-to-background ratio and a large significance value. The results demonstrate the reliability and high efficiency of the missing mass method in the reconstruction of both charged mother particles and their neutral daughter particles. Being an integral part of the KF Particle Finder package, now the main approach for reconstruction and analysis of short-lived particles in the STAR experiment, the missing mass method will continue to be used for the physics analysis in online and offline modes.
The high quality of the results of the express data analysis has led to their status as preliminary physics results with the right to present them at international physics conferences and meetings on behalf of the STAR Collaboration.
Correction to: Computational Economics https://doi.org/10.1007/s10614-020-10061-x
The original publication has been updated. In the original publication of this article, under the Introduction heading section, the corrections to the second paragraph’s inline equation were not incorporated. The author’s additional corrections have also been incorporated. The publisher apologizes for the error made during production.
With open banking, consumers take greater control over their own financial data and share it at their discretion. Using a rich set of loan application data from the largest German FinTech lender in consumer credit, this paper studies what characterizes borrowers who share data and assesses its impact on loan application outcomes. I show that riskier borrowers share data more readily, which subsequently leads to an increase in the probability of loan approval and a reduction in interest rates. The effects hold across all credit risk profiles but are the most pronounced for borrowers with lower credit scores (a higher increase in loan approval rate) and higher credit scores (a larger reduction in interest rate). I also find that standard variables used in credit scoring explain substantially less variation in loan application outcomes when customers share data. Overall, these findings suggest that open banking improves financial inclusion, and also provide policy implications for regulators engaged in the adoption or extension of open banking policies.
Statistical shape models learn to capture the most characteristic geometric variations of anatomical structures given samples from their population. Accordingly, shape models have become an essential tool for many medical applications and are used in, for example, shape generation, reconstruction, and classification tasks. However, established statistical shape models require precomputed dense correspondence between shapes, often lack robustness, and ignore the global surface topology. This thesis presents a novel neural flow-based shape model that does not require any precomputed correspondence. The proposed model relies on continuous flows of a neural ordinary differential equation to model shapes as deformations of a template. To increase the expressivity of the neural flow and disentangle global, low-frequency deformations from the generation of local, high- frequency details, we propose to apply a hierarchy of flows. We evaluate the performance of our model on two anatomical structures, liver, and distal femur. Our model outperforms state-of-the-art methods in providing an expressive and robust shape prior, as indicated by its generalization ability and specificity. More so, we demonstrate the effectiveness of our shape model on shape reconstruction tasks and find anatomically plausible solutions. Finally, we assess the quality of the emerging shape representation in an unsupervised setting and discriminate healthy from pathological shapes.
Debate topic expansion
(2022)
Given a debate topic, it is often to make an expansion of the topic, the reasons can be the followings: (1) The scope of the debate topic is too shallow and we eager to discuss more. (2) A debate topic is sometimes related to the others and the discussion will not be complete when we do not discuss the others as well. (3) We may want to discuss the particular concept or the core the debate topic. It's thus meaningful to build a model in order to find the expansions of the topics.
IBM Research Team has proposed a method to expand the boundary and find the expansion topics of the given debate topics in 2019. There are two types of topic expansions in their paper, consistent and contrastive expansions. We focus on the consistent expansions. Consistent expansions are defined as the expansions that expand our topics in a positive way or at least neutral.
The main objective of this paper is to follow and examine the steps of IBM Research Team's idea and since the original discusses the model in english, we would like to implement a topic expansion model with 7 steps, including pattern extraction, filtering, training, etc, in another language (german) using translator and compare the result between different models to propose the final german model at the end.
Das Projekt anan ist ein Werkzeug zur Fehlersuche in verteilten Hochleistungsrechnern. Die Neuheit des Beitrags besteht darin, dass die bekannten Methoden, die bereits erfolgreich zum Debuggen von Soft- und Hardware eingesetzt werden, auf Hochleistungs-Rechnen übertragen worden sind. Im Rahmen der vorliegenden Arbeit wurde ein Werkzeug namens anan implementiert, das bei der Fehlersuche hilft. Außerdem kann es als dynamischeres Monitoring eingesetzt werden. Beide Einsatzzwecke sind
getestet worden.
Das Werkzeug besteht aus zwei Teilen:
1. aus einem Teil namens anan, der interaktiv vom Nutzer bedient wird
2. und aus einem Teil namens anand, der automatisiert die verlangten Messwerte erhebt und nötigenfalls Befehle ausführt.
Der Teil anan führt Sensoren aus — kleine mustergesteuerte Algorithmen —, deren Ergebnisse per anan zusammengeführt werden. In erster Näherung lässt anan sich als Monitoring beschreiben, welches (1) schnell umkonfiguriert werden (2) komplexere Werte messen kann, die über Korrelationen einfacher Zeitreihen hinausgehen.
With free delivery of products virtually being a standard in E-commerce, product returns pose a major challenge for online retailers and society. For retailers, product returns involve significant transportation, labor, disposal, and administrative costs. From a societal perspective, product returns contribute to greenhouse gas emissions and packaging disposal and are often a waste of natural resources. Therefore, reducing product returns has become a key challenge. This paper develops and validates a novel smart green nudging approach to tackle the problem of product returns during customers’ online shopping processes. We combine a green nudge with a novel data enrichment strategy and a modern causal machine learning method. We first run a large-scale randomized field experiment in the online shop of a German fashion retailer to test the efficacy of a novel green nudge. Subsequently, we fuse the data from about 50,000 customers with publicly-available aggregate data to create what we call enriched digital footprints and train a causal machine learning system capable of optimizing the administration of the green nudge. We report two main findings: First, our field study shows that the large-scale deployment of a simple, low-cost green nudge can significantly reduce product returns while increasing retailer profits. Second, we show how a causal machine learning system trained on the enriched digital footprint can amplify the effectiveness of the green nudge by “smartly” administering it only to certain types of customers. Overall, this paper demonstrates how combining a low-cost marketing instrument, a privacy-preserving data enrichment strategy, and a causal machine learning method can create a win-win situation from both an environmental and economic perspective by simultaneously reducing product returns and increasing retailers’ profits.
Meat adulteration is a global problem which undermines market fairness and harms people with allergies or certain religious beliefs. In this study, a novel framework in which a one-dimensional convolutional neural network (1DCNN) serves as a backbone and a random forest regressor (RFR) serves as a regressor, named 1DCNN-RFR, is proposed for the quantitative detection of beef adulterated with pork using electronic nose (E-nose) data. The 1DCNN backbone extracted a sufficient number of features from a multichannel input matrix converted from the raw E-nose data. The RFR improved the regression performance due to its strong prediction ability. The effectiveness of the 1DCNN-RFR framework was verified by comparing it with four other models (support vector regression model (SVR), RFR, backpropagation neural network (BPNN), and 1DCNN). The proposed 1DCNN-RFR framework performed best in the quantitative detection of beef adulterated with pork. This study indicated that the proposed 1DCNN-RFR framework could be used as an effective tool for the quantitative detection of meat adulteration.
Although everyone is familiar with using algorithms on a daily basis, formulating, understanding and analysing them rigorously has been (and will remain) a challenging task for decades. Therefore, one way of making steps towards their understanding is the formulation of models that are portraying reality, but also remain easy to analyse. In this thesis we take a step towards this way by analyzing one particular problem, the so-called group testing problem. R. Dorfman introduced the problem in 1943. We assume a large population and in this population we find a infected group of individuals. Instead of testing everybody individually, we can test group (for instance by mixing blood samples). In this thesis we look for the minimum number of tests needed such that we can say something meaningful about the infection status. Furthermore we assume various versions of this problem to analyze at what point and why this problem is hard, easy or impossible to solve.
Im Fachbereich der Computerlinguistik ist die automatische Generierung von Szenen aus, in natürlicher Sprache verfassten, Text seit bereits vielen Jahrzehnten ein wichtiger Bestandteil der Forschung, welche in der "Kunst", "Lehre" und "Robotik" Verwendung finden. Mit Hilfe von neuen Technologien im Bereich der Künstlichen Intelligenzen (KI), werden neue Entwicklungen möglich, welche diese Generierungen vereinfachen, allerdings auch undurchsichtige interne vom Modell getroffene Entscheidungen fördern.
Ziel der vorgeschlagenen Lösung „ARES: Annotation von Relationen und Eigenschaften zur Szenengenerierung“ ist es, ein modulares System zu entwerfen, wobei einzelne Prozesse für den Benutzer verständlich bleiben. Außerdem sollen Möglichkeiten geboten werden, neue Entitäten und Relationen, welche über die Textanalyse bereitgestellt werden, auch in die Szenengenerierung im dreidimensionalen Raum einzupflegen, ohne dass hierfür Code zwingend notwendig wird.
Der Fokus liegt auf der syntaktisch korrekten Darstellung der Elemente im Raum. Dagegen lässt sich die semantische Korrektheit durch weitere manuelle Anpassungen, welche für spätere Generierungen gespeichert werden erhöhen. Letztlich soll die Menge der zur Darstellung benötigten Annotationen möglichst gering bleiben und neue szenenbezogene Annotationen durch die implementierten Annotationstools hinzugefügt werden.
Modeling long-term neuronal dynamics may require running long-lasting simulations. Such simulations are computationally expensive, and therefore it is advantageous to use simplified models that sufficiently reproduce the real neuronal properties. Reducing the complexity of the neuronal dendritic tree is one option. Therefore, we have developed a new reduced-morphology model of the rat CA1 pyramidal cell which retains major dendritic branch classes. To validate our model with experimental data, we used HippoUnit, a recently established standardized test suite for CA1 pyramidal cell models. The HippoUnit allowed us to systematically evaluate the somatic and dendritic properties of the model and compare them to models publicly available in the ModelDB database. Our model reproduced (1) somatic spiking properties, (2) somatic depolarization block, (3) EPSP attenuation, (4) action potential backpropagation, and (5) synaptic integration at oblique dendrites of CA1 neurons. The overall performance of the model in these tests achieved higher biological accuracy compared to other tested models. We conclude that, due to its realistic biophysics and low morphological complexity, our model captures key physiological features of CA1 pyramidal neurons and shortens computational time, respectively. Thus, the validated reduced-morphology model can be used for computationally demanding simulations as a substitute for more complex models.
Korrektur zu: Höllbacher, S., Wittum, G. Correction to: A sharp interface method using enriched finite elements for elliptic interface problems. Numer. Math. 147, 783 (2021). DOI: 10.1007/s00211-021-01180-0.
We present an immersed boundary method for the solution of elliptic interface problems with discontinuous coefficients which provides a second-order approximation of the solution. The proposed method can be categorised as an extended or enriched finite element method. In contrast to other extended FEM approaches, the new shape functions get projected in order to satisfy the Kronecker-delta property with respect to the interface. The resulting combination of projection and restriction was already derived in Höllbacher and Wittum (TBA, 2019a) for application to particulate flows. The crucial benefits are the preservation of the symmetry and positive definiteness of the continuous bilinear operator. Besides, no additional stabilisation terms are necessary. Furthermore, since our enrichment can be interpreted as adaptive mesh refinement, the standard integration schemes can be applied on the cut elements. Finally, small cut elements do not impair the condition of the scheme and we propose a simple procedure to ensure good conditioning independent of the location of the interface. The stability and convergence of the solution will be proven and the numerical tests demonstrate optimal order of convergence.
In der Arbeit wird das Certainty-Tool, eine Erweiterung für den Unity-basierten Teil des Stolperwege Projektes, vorgestellt. Dieses verfolgt die Idee des VAnnotatoR weiter und erlaubt die Visualisierung von informationeller Ungewissheit der im Stolperwege-Praktikum digital rekonstruierten Gebäude. Dabei inkorporiert das Tool das Konzept hinter BIM (Building Information Modelling), eine neuartige Methode der Planung in der AEC-Branche, welches ein Selbstbewusstsein von Informationen für Teile eines Gebäudes ermöglicht. Dabei werden im Certainty-Tool Stufen der informationellen Ungewissheit entwickelt und diese auf Teile des Gebäudes zugewiesen. Das Tool wird anhand einer digitalen Rekonstruktion des zerstörten Rothschild-Palais vorgestellt. Des Weiteren wurde eine Evaluation basierend auf der Usability Metric for User Experience durchgeführt und weiterführende Entwicklungen und Verbesserungen des Tools diskutiert.
Bei der Bekleidungsmodellierung geht es um den Entwurf von Bekleidung von Personen, die beispielsweise in Szenen dargestellt werden können. Dabei stützt sich der Entwurf auf Informationen aus einer Datengrundlage. Die Darstellung von Szenen, in denen Personen dargestellt werden, stellt sich grundsätzlich als Zusammenspiel komplexer Teilaspekte dar. Dabei wird die Nachvollziehbarkeit einer modellierten Szene oder modellierter Avatare im Auge des Betrachters ganz wesentlich durch den Faktor passend gewählter Kleidung bestimmt.
In dieser Arbeit werden Ansätze und Verfahren vorgestellt, die zur Bekleidungsmodellierung auf Grundlage von Textdokumenten basieren. Dafür werden Möglichkeiten erörtert, die es erlauben Informationen aus Texten zu extrahieren und für die Modellierung einzusetzen.
Zur Bearbeitung der Aufgabenstellung wird zunächst ein aus dem Machine Learning bekanntes kontextuelles Modell hinsichtlich einer Mehrklassen-Klassifizierung trainiert und angewendet. Daraufhin wird die Erstellung einer eigenen Wissensressource, die sich auf textlicher Ebene mit dem Thema der Bekleidung auseinandersetzt, aufgebaut und mit zahlreichen Informationen aus bereits bestehenden Ressourcen popularisiert. Die neue Ressource wird in Form einer Graphdatenbank entworfen. Dabei werden Relationen zwischen den einzelnen Elementen mithilfe von statischen Modellen sowie einem kontextuellen Modell, dem BERT-Modell, erstellt. Schließlich wird auf Grundlage der entwickelten Graphdatenbank ein in der Programmiersprache Python entwickeltes Programm vorgestellt, dass Eingabetexte unter Hinzunahme der Informationen und Relationen innerhalb der Graphdatenbank verarbeitet und Kleidungsstücke detektiert.
Nach der theoretischen Aufarbeitung der entwickelten Ansätze werden die daraus resultierenden Ergebnisse diskutiert und bestehende Problematiken bei der Bearbeitung der Aufgabenstellung angesprochen. Abschließend wird die Arbeit zusammengefasst und Anregungen für die weitere Bearbeitung dieser Thematik vorgestellt.
Reactive oxygen species are a class of naturally occurring, highly reactive molecules that change the structure and function of macromolecules. This can often lead to irreversible intracellular damage. Conversely, they can also cause reversible changes through post-translational modification of proteins which are utilized in the cell for signaling. Most of these modifications occur on specific cysteines. Which structural and physicochemical features contribute to the sensitivity of cysteines to redox modification is currently unclear. Here, I investigated the in uence of protein structural and sequence features on the modifiability of proteins and specific cysteines therein using statistical and machine learning methods. I found several strong structural predictors for redox modification, such as a higher accessibility to the cytosol and a high number of positively charged amino acids in the close vicinity. I detected a high frequency of other post-translational modifications, such as phosphorylation and ubiquitination, near modified cysteines. Distribution of secondary structure elements appears to play a major role in the modifiability of proteins. Utilizing these features, I created models to predict the presence of redox modifiable cysteines in proteins, including human mitochondrial complex I, NKG2E natural killer cell receptors and proximal tubule cell proteins, and compared some of these predictions to earlier experimental results.
This thesis concerns three specific constraint satisfaction problems: the k-SAT problem, random linear equations and the Potts model. We investigated a phenomenon called replica symmetry, its consequences and its limitation. For the $k$-SAT problem, we were able to show that replica symmetry holds up to a threshold $d^{*}$. However, after another critical threshold $d^{**}$, we discovered that replica symmetry could not hold anymore, which enabled us to establish the existence of a replica symmetry breaking region. For the random linear problem, a peculiar phenomenon occurs. We observed that a more robust version of replica symmetry (strong replica symmetry) holds up to a threshold $d=e$ and ceases to hold after. This phenomenon is linked to the fact that before the threshold $d=e$, the fraction of frozen variables, i.e. variable forced to take the same value in all solutions, is concentrated around a deterministic value but vacillates between two values with equal probability for $d>e$. Lastly, for the Potts model, we show that a phenomenon called metastability occurs. The latter phenomenon can be understood as a consequence of trivial replica symmetry breaking scheme. This metastability phenomenon further produces slow mixing results for two famous Markov chains, the Glauber and the Swendsen-Wang dynamics.
The relevant field of interest in High Energy Physics experiments is shifting to searching and studying extremely rare particles and phenomena. The search for rare probes requires an increase in the number of available statistics by increasing the particle interaction rate. The structure of the events also becomes more complicated, the multiplicity of particles in each event increases, and a pileup appears. Due to technical limitations, such data flow becomes impossible to store fully on available storage devices. The solution to the problem is the correct triggering of events and real-time data processing.
In this work, the issue of accelerating and improving the algorithms for reconstruction of the charged particles' trajectories based on the Cellular Automaton in the STAR experiment is considered to implement them for track reconstruction in real-time within the High-Level Trigger. This is an important step in the preparation of the CBM experiment as part of the FAIR Phase-0 program. The study of online data processing methods in real conditions at similar interaction energies allows us to study this process and determine the possible weaknesses of the approach.
Two versions of the Cellular Automaton based track reconstruction are discussed, which are used, depending on the detecting systems' features. HFT~CA Track Finder, similar to the tracking algorithm of the CBM experiment, has been accelerated by several hundred times, using both algorithm optimization and data-level parallelism. TPC~CA Track Finder has been upgraded to improve the reconstruction quality while maintaining high calculation speed. The algorithm was tuned to work with the new iTPC geometry and provided an additional module for very low momentum track reconstruction.
The improved track reconstruction algorithm for the TPC detector in the STAR experiment was included in the HLT reconstruction chain and successfully tested in the express production for the online real data analysis. This made it possible to obtain important physical results during the experiment runtime without the full offline data processing. The tracker is also being prepared for integration into a standard offline data processing chain, after which it will become the basic track search algorithm in the STAR experiment.
Die allgemein steigende Komplexität technischer Systeme macht sich auch in eingebetteten Systemen bemerkbar. Außerdem schrumpfen die Strukturgrößen der eingesetzten Komponenten, was wiederum die Auftrittswahrscheinlichkeit verschiedener Effekte erhöht, die zu Fehlern und Ausfällen dieser Komponenten und damit der Gesamtsysteme führen können. Da in vielen Anwendungsbereichen ferner Sicherheitsanforderungen eingehalten werden müssen, sind zur Gewährleistung der Zuverlässigkeit flexible Redundanzkonzepte nötig.
Ein Forschungsgebiet, das sich mit Methoden zur Beherrschung der Systemkomplexität befasst, ist das Organic Computing. In dessen Rahmen werden Konzepte erforscht, um in natürlichen Systemen beobachtbare Eigenschaften und Organisationsprinzipien auf technische Systeme zu übertragen. Hierbei sind insbesondere sogenannte Selbst-X-Eigenschaften wie Selbstorganisation, -konfiguration und -heilung von Bedeutung.
Eine konkrete Ausprägung dieses Forschungszweigs ist das künstliche Hormonsystem (artificial hormone system, AHS). Hierbei handelt es sich um eine Middleware für verteilte Systeme, welche es ermöglicht, die Tasks des Systems selbstständig auf seine Prozessorelemente (PEs) zu verteilen und insbesondere Ausfälle einzelner Tasks oder ganzer PEs automatisch zu kompensieren, indem die betroffenen Tasks auf andere PEs migriert werden. Hierbei existiert keine zentrale Instanz, welche die Taskverteilung steuert und somit einen Single-Point-of-Failure darstellen könnte. Entsprechend kann das AHS aufgrund seiner automatischen (Re)konfiguration der Tasks als selbstkonfigurierend und selbstheilend bezeichnet werden, was insbesondere die Zuverlässigkeit des realisierten Systems erhöht. Die Dauer der Selbstkonfiguration und Selbstheilung unterliegt zudem harten Zeitschranken, was den Einsatz des AHS auch in Echtzeitsystemen erlaubt.
Das AHS nimmt jedoch an, dass alle Tasks gleichwertig sind, zudem werden alle Tasks beim Systemstart in einer zufälligen Reihenfolge auf die einzelnen PEs verteilt. Häufig sind die in einem System auszuführenden Tasks jedoch für das Gesamtsystem von unterschiedlicher Wichtigkeit oder müssen gar in einer bestimmten Reihenfolge gestartet werden.
Um den genannten Eigenschaften Rechnung zu tragen, liefert diese Dissertation gegenüber dem aktuellen Stand der Forschung folgende Beiträge:
Zunächst werden die bisher bekannten Zeitschranken des AHS genauer betrachtet und verfeinert.
Anschließend wird das AHS durch die Einführung von Zuteilungsprioritäten erweitert: Mithilfe dieser Prioritäten kann eine Reihenfolge definiert werden, in welcher die Tasks beim Start des Systems auf die PEs verteilt beziehungsweise in welcher betroffene Tasks nach einem Ausfall auf andere PEs migriert werden.
Die Zeitschranken dieser AHS-Erweiterung werden im Detail analysiert.
Durch die Priorisierung von Tasks ist es möglich, implizit Teilmengen von Tasks zu definieren, die ausgeführt werden sollen, falls die Rechenkapazitäten des Systems nach einer bestimmten Anzahl von PE-Ausfällen nicht mehr ausreichen, um alle Tasks auszuführen: Die im Rahmen dieser Dissertation entwickelten Erweiterungen erlauben es in solchen Überlastsituationen, das System automatisch und kontrolliert zu degradieren, sodass die wichtigsten Systemfunktionalitäten lauffähig bleiben.
Überlastsituationen werden daher im Detail betrachtet und analysiert. In solchen müssen gegebenenfalls Tasks niedriger Priorität gestoppt werden, um auf den funktionsfähig verbleibenden PEs hinreichend viel Rechenkapazität zu schaffen, um Tasks höherer Priorität ausführen zu können und das System so in einen wohldefinierten Zustand zu überführen. Die Entscheidung, in welcher Reihenfolge hierbei Tasks gestoppt werden, wird von einer Task-Dropping-Strategie getroffen, die entsprechend einen großen Einfluss auf die Dauer einer solchen Selbstheilung nimmt.
Es werden zwei verschiedene Task-Dropping-Strategien entwickelt und im Detail analysiert: die naive Task-Dropping-Strategie, welche alle niedrigprioren Tasks auf einmal stoppt, sowie das Eager Task Dropping, das in mehreren Phasen jeweils höchstens eine Task pro PE stoppt. Im Vergleich zeigt sich, dass von letzterem fast immer weniger Tasks gestoppt werden als von der naiven Strategie, was einen deutlich schnelleren Abschluss der Selbstheilung ermöglicht. Lediglich in wenigen Sonderfällen ist die naive Strategie überlegen.
Es wird detailliert gezeigt, dass die entwickelte AHS-Erweiterung auch in Überlastsituationen die Einhaltung bestimmter harter Zeitschranken garantieren kann, was den Einsatz des erweiterten AHS in Echtzeitsystemen erlaubt.
Alle theoretisch hergeleiteten Zeitschranken werden durch umfassende Evaluationen vollumfänglich bestätigt.
Abschließend wird das erweiterte, prioritätsbasierten AHS mit verschiedenen verwandten Konzepten verglichen, um dessen Vorteile gegenüber dem Stand der Forschung herauszuarbeiten sowie zukünftige vertiefende Forschung zu motivieren.
Efficient algorithms for object recognition are crucial for the newly robotics and computer vision applications that demand real-time and on-line methods. Some examples are autonomous systems, navigating robots, autonomous driving. In this work, we focus on efficient semantic segmentation, which is the problem of labeling each pixel of an image with a semantic class.
Our aim is to speed-up all of the parts of the semantic segmentation pipeline. We also aim at delivering a labeling solution on a time budget, that can be decided on-the-fly. For this purpose, we analyze all the components of the semantic segmentation pipeline, and identify the computational bottleneck of each of them. The different components of the pipeline are over-segmenting the image with local regions, extracting features and classify the local regions, and the final inference of the image labeling with semantic classes. We focus on each of these steps.
First, we introduce a new superpixel algorithm to over-segment the image. Our superpixel method runs in real-time and can deliver a solution at any time budget. Then, for feature extraction, we focus on the framework that computes descriptors and encodes them, followed by a pooling step. We see that the encoding step is the bottleneck, for computational efficiency and performance. We present a novel assignment-based encoding formulation, that allows for the design of a new, very efficient, encoding. Finally, the image labeling output is obtained modeling the dependencies with a Conditional Random Field (CRF). In semantic image segmentation, the computational cost of instantiating the potentials is much higher than MAP inference. We introduce Active MAP inference to on-the-fly select a subset of potentials to be instantiated in the energy function, leaving the rest as unknown, and to estimate the MAP labeling from such incomplete energy function.
We perform experiments on all proposed methods for the different parts of the semantic segmentation pipeline. We show that our superpixel extraction achieves higher accuracy than state-of-the-art on standard superpixel benchmark, while it runs in real-time. We test our feature encoding on standard image classification and segmentation benchmarks, and we show that our method achieves competitive results with the state-of-the-art, and requires less time and memory. Finally, results for semantic segmentation benchmark show that Active MAP inference achieves similar levels of accuracy but with major efficiency gains.
When performing transfer learning in Computer Vision, normally a pretrained model (source model) that is trained on a specific task and a large dataset like ImageNet is used. The learned representation of that source model is then used to perform a transfer to a target task. Performing transfer learning in this way had a great impact on Computer Vision, because it worked seamlessly, especially on tasks that are related to each other. Current research topics have investigated the relationship between different tasks and their impact on transfer learning by developing similarity methods. These similarity methods have in common, to do transfer learning without actually doing transfer learning in the first place but rather by predicting transfer learning rankings so that the best possible source model can be selected from a range of different source models. However, these methods have focused only on singlesource transfers and have not paid attention to multi-source transfers. Multi-source transfers promise even better results than single-source transfers as they combine information from multiple source tasks, all of which are useful to the target task. We fill this gap and propose a many-to-one task similarity method called MOTS that predicts both, single-source transfers and multi-source transfers to a specific target task. We do that by using linear regression and the source representations of the source models to predict the target representation. We show that we achieve at least results on par with related state-of-the-art methods when only focusing on singlesource transfers using the Pascal VOC and Taskonomy benchmark. We show that we even outperform all of them when using single and multi-source transfers together (0.9 vs. 0.8) on the Taskonomy benchmark. We additionally investigate the performance of MOTS in conjunction with a multi-task learning architecture. The task-decoder heads of a multi-task learning architecture are used in different variations to do multi-source transfers since it promises efficiency over multiple singletask architectures and incurs less computational cost. Results show that our proposed method accurately predicts transfer learning rankings on the NYUD dataset and even shows the best transfer learning results always being achieved when using more than one source task. Additionally, it is further examined that even just using one task-decoder head from the multi-task learning architecture promises better transfer learning results, than using a single-task architecture for the same task, which is due to the shared information from different tasks in the multi-task learning architecture in previous layers. Since the MOTS rankings for selecting the MTI-Net task-decoder head with the highest transfer learning performance were very accurate for the NYUD but not satisfying for the Pascal VOC dataset, further experiments need to varify the generalizability of MOTS rankings for the selection of the optimal task-decoder head from a multi-task architecture.
The archaeological data dealt with in our database solution Antike Fundmünzen in Europa (AFE), which records finds of ancient coins, is entered by humans. Based on the Linked Open Data (LOD) approach, we link our data to Nomisma.org concepts, as well as to other resources like Online Coins of the Roman Empire (OCRE). Since information such as denomination, material, etc. is recorded for each single coin, this information should be identical for coins of the same type. Unfortunately, this is not always the case, mostly due to human errors. Based on rules that we implemented, we were able to make use of this redundant information in order to detect possible errors within AFE, and were even able to correct errors in Nomimsa.org. However, the approach had the weakness that it was necessary to transform the data into an internal data model. In a second step, we therefore developed our rules within the Linked Open Data world. The rules can now be applied to datasets following the Nomisma. org modelling approach, as we demonstrated with data held by Corpus Nummorum Thracorum (CNT). We believe that the use of methods like this to increase the data quality of individual databases, as well as across different data sources and up to the higher levels of OCRE and Nomisma.org, is mandatory in order to increase trust in them.
Solving High-Dimensional Dynamic Portfolio Choice Models with Hierarchical B-Splines on Sparse Grids
(2021)
Discrete time dynamic programming to solve dynamic portfolio choice models has three immanent issues: firstly, the curse of dimensionality prohibits more than a handful of continuous states. Secondly, in higher dimensions, even regular sparse grid discretizations need too many grid points for sufficiently accurate approximations of the value function. Thirdly, the models usually require continuous control variables, and hence gradient-based optimization with smooth approximations of the value function is necessary to obtain accurate solutions to the optimization problem. For the first time, we enable accurate and fast numerical solutions with gradient-based optimization while still allowing for spatial adaptivity using hierarchical B-splines on sparse grids. When compared to the standard linear bases on sparse grids or finite difference approximations of the gradient, our approach saves an order of magnitude in total computational complexity for a representative dynamic portfolio choice model with varying state space dimensionality, stochastic sample space, and choice variables.
Gegenstände und Gesichter können Computer schon recht gut erkennen, auch dass sich etwas bewegt und in welche Richtung. Schwierigkeiten bereiten der Künstlichen Intelligenz aber noch zu erfassen, um welche Art von Bewegungen es sich handelt. Das lernen Computer jetzt im Labor von Prof. Hilde Kühne an der Goethe-Universität.
The present research in high energy physics as well as in the nuclear physics requires the use of more powerful and complex particle accelerators to provide high luminosity, high intensity, and high brightness beams to experiments. With the increased technological complexity of accelerators, meeting the demand of experimenters necessitates a blend of accelerator physics with technology. The problem becomes severe when optimization of beam quality has to be provided in accelerator systems with thousands of free parameters including strengths of quadrupoles, sextupoles, RF voltages, etc. Machine learning methods and concepts of artificial intelligence are considered in various industry and scientific branches, and recently, these methods are used in high energy physics mainly for experiments data analysis.
In Accelerator Physics the machine learning approach has not found a wide application yet, and in general the use of these methods is carried out without a deep understanding on their effectiveness with respect to more traditional schemes or other alternative approaches. The purpose of this PhD research is to investigate the methods of machine learning applied to accelerator optimization, accelerator control and in particular on optics measurements and corrections. Optics correction, maximization of acceptance, and simultaneous control of various accelerator components such as focusing magnets is a typical accelerator scenario. The effectiven- ess of machine learning methods in a complex system such as the Large Hadron Collider, which beam dynamics exhibits nonlinear response to machine settings is the core of the study. This work presents successful application of several machine learning techniques such as clustering, decision trees, linear multivariate models and neural networks to beam optics measurements and corrections at the LHC, providing the guidelines for incorporation of machine learning techniques into accelerator operation and discussing future opportunities and potential work in this field.
Reproducible annotations
(2022)
This bachelor thesis presents a software solution which implements reproducible annotations in the context of the UIMA framework. This is achieved by creating an automated containerization of arbitrary analysis engines and annotating every analysis engine configuration in the processed CAS document. Any CAS document created by this solution is self sufficient and able to reproduce the exact environment under which it was created.
A review of the state-of-the art software in the field of UIMA reveals that there are many implementations trying to increase reproducibility for a given application relying on UIMA, but no publication trying to increase the reproducibility of UIMA itself. This thesis improves upon that technological gap and provides a throughout analysis at the end which shows a negligible overhead in memory consumption, but a significant performance regression depending on the complexity of the analysis engine which was examined.
Ein aktuelles Forschungsthema ist die automatische Generierung von 3D-Szenen ausgehend von Beschreibungen in natürlicher Sprache. S.g. Text2Scene-Anwendungen sollen Objekte und räumliche Relationen in einer Texteingabe identifizieren und mit 3D-Modellen eine visuelle Repräsentation der Beschreibung konstruieren. Bisherige Ansätze kombinieren eine
stichwortbasierte Erkennung von explizit gemachten Angaben mit vorher gelerntem Allgemeinwissen über die sinnvolle Anordnung von Objekten. Den Anwendungen fehlt jedoch ein tiefergehendes Verständnis von räumlicher Sprache.
Mit dem Annotationsschema ISOSpace können Texte mit detaillierten räumlichen Informationen angereichert und so für NLP-Anwendungen verständlicher gemacht werden. Bereits in einer früheren Arbeit wurde der SemAF-Annotator zum Erstellen von ISOSpaceAnnotationen als Modul für den TextAnnotator entwickelt. In dieser Arbeit wurde der SemAF-Annotator zusätzlich um eine Funktionalität zur Szenenerstellung erweitert: Benutzer können einzelnen Wörtern in der Weboberfläche des TextAnnotators Objekte aus dem ShapeNet Datensatz zuordnen und diese in einer zweidimensionalen Darstellung einer Szene räumlich anordnen. Trotz einiger Einschränkungen durch die fehlende dritte Dimension lassen sich in vielen Fällen gute Ergebnisse erzielen. Die auf diese Weise erzeugten Szenen sollen später in Kombination mit den ISOSpace-Annotionen verwendet werden, um Text2SceneAnwendungen zu entwickeln, die ein umfassenderes räumliches Verständnis aufweisen.
Kleinere Nebenaufgaben dieser Arbeit waren die Erweiterung des SemAF-Annotators um zusätzliche Annotationstypen sowie diverse Nachbesserungen der bereits bestehenden Funktionalität zur ISOSpace Annotation.
The recognition of pharmacological substances, compounds and proteins is an essential preliminary work for the recognition of relations between chemicals and other biomedically relevant units. In this paper, we describe an approach to Task 1 of the PharmaCoNER Challenge, which involves the recognition of mentions of chemicals and drugs in Spanish medical texts. We train a state-of-the-art BiLSTM-CRF sequence tagger with stacked Pooled Contextualized Embeddings, word and sub-word embeddings using the open-source framework FLAIR. We present a new corpus composed of articles and papers from Spanish health science journals, termed the Spanish Health Corpus, and use it to train domain-specific embeddings which we incorporate in our model training. We achieve a result of 89.76% F1-score using pre-trained embeddings and are able to improve these results to 90.52% F1-score using specialized embeddings.
Despite the great importance of the Latin language in the past, there are relatively few resources available today to develop modern NLP tools for this language. Therefore, the EvaLatin Shared Task for Lemmatization and Part-of-Speech (POS) tagging was published in the LT4HALA workshop. In our work, we dealt with the second EvaLatin task, that is, POS tagging. Since most of the available Latin word embeddings were trained on either few or inaccurate data, we trained several embeddings on better data in the first step. Based on these embeddings, we trained several state-of-the-art taggers and used them as input for an ensemble classifier called LSTMVoter. We were able to achieve the best results for both the cross-genre and the cross-time task (90.64% and 87.00%) without using additional annotated data (closed modality). In the meantime, we further improved the system and achieved even better results (96.91% on classical, 90.87% on cross-genre and 87.35% on cross-time).
Biodiversity information is contained in countless digitized and unprocessed scholarly texts. Although automated extraction of these data has been gaining momentum for years, there are still innumerable text sources that are poorly accessible and require a more advanced range of methods to extract relevant information. To improve the access to semantic biodiversity information, we have launched the BIOfid project (www.biofid.de) and have developed a portal to access the semantics of German language biodiversity texts, mainly from the 19th and 20th century. However, to make such a portal work, a couple of methods had to be developed or adapted first. In particular, text-technological information extraction methods were needed, which extract the required information from the texts. Such methods draw on machine learning techniques, which in turn are trained by learning data. To this end, among others, we gathered the BIOfid text corpus, which is a cooperatively built resource, developed by biologists, text technologists, and linguists. A special feature of BIOfid is its multiple annotation approach, which takes into account both general and biology-specific classifications, and by this means goes beyond previous, typically taxon- or ontology-driven proper name detection. We describe the design decisions and the genuine Annotation Hub Framework underlying the BIOfid annotations and present agreement results. The tools used to create the annotations are introduced, and the use of the data in the semantic portal is described. Finally, some general lessons, in particular with multiple annotation projects, are drawn.
Are nearby places (e.g., cities) described by related words? In this article, we transfer this research question in the field of lexical encoding of geographic information onto the level of intertextuality. To this end, we explore Volunteered Geographic Information (VGI) to model texts addressing places at the level of cities or regions with the help of so-called topic networks. This is done to examine how language encodes and networks geographic information on the aboutness level of texts. Our hypothesis is that the networked thematizations of places are similar, regardless of their distances and the underlying communities of authors. To investigate this, we introduce Multiplex Topic Networks (MTN), which we automatically derive from Linguistic Multilayer Networks (LMN) as a novel model, especially of thematic networking in text corpora. Our study shows a Zipfian organization of the thematic universe in which geographical places (especially cities) are located in online communication. We interpret this finding in the context of cognitive maps, a notion which we extend by so-called thematic maps. According to our interpretation of this finding, the organization of thematic maps as part of cognitive maps results from a tendency of authors to generate shareable content that ensures the continued existence of the underlying media. We test our hypothesis by example of special wikis and extracts of Wikipedia. In this way, we come to the conclusion that geographical places, whether close to each other or not, are located in neighboring semantic places that span similar subnetworks in the topic universe.
The annotation of texts and other material in the field of digital humanities and Natural Language Processing (NLP) is a common task of research projects. At the same time, the annotation of corpora is certainly the most time- and cost-intensive component in research projects and often requires a high level of expertise according to the research interest. However, for the annotation of texts, a wide range of tools is available, both for automatic and manual annotation. Since the automatic pre-processing methods are not error-free and there is an increasing demand for the generation of training data, also with regard to machine learning, suitable annotation tools are required. This paper defines criteria of flexibility and efficiency of complex annotations for the assessment of existing annotation tools. To extend this list of tools, the paper describes TextAnnotator, a browser-based, multi-annotation system, which has been developed to perform platform-independent multimodal annotations and annotate complex textual structures. The paper illustrates the current state of development of TextAnnotator and demonstrates its ability to evaluate annotation quality (inter-annotator agreement) at runtime. In addition, it will be shown how annotations of different users can be performed simultaneously and collaboratively on the same document from different platforms using UIMA as the basis for annotation.
Wir betrachten Algorithmen für strategische Kommunikation mit Commitment Power zwischen zwei rationalen Parteien mit eigenen Interessen. Wenn eine Partei Commitment Power hat, so legt sie sich auf eine Handlungsstrategie fest und veröffentlicht diese und kann nicht mehr davon abweichen.
Beide Parteien haben Grundinformation über den Zustand der Welt. Die erste Partei (S) hat die Möglichkeit, diesen direkt zu beobachten. Die zweite Partei (R) trifft jedoch eine Entscheidung durch die Wahl einer von n Aktionen mit für sie unbekanntem Typ. Dieser Typ bestimmt die möglicherweise verschiedenen, nicht-negativen Nutzwerte für S und R. Durch das Senden von Signalen versucht S, die Wahl von R zu beeinflussen. Wir betrachten zwei Grundszenarien: Bayesian Persuasion und Delegated Search.
In Bayesian Persuasion besitzt S Commitment Power. Hier legt sich S sich auf ein Signalschema φ fest und teilt dieses R mit. Es beschreibt, welches Signal S in welcher Situation sendet. Erst danach erfährt S den wahren Zustand der Welt. Nach Erhalt der durch φ bestimmten Signale wählt R eine der Aktionen. Das Wissen um φ erlaubt R die Annahmen über den Zustand der Welt in Abhängigkeit von den empfangenen Signalen zu aktualisieren. Dies muss S für das Design von φ berücksichtigen, denn R wird Empfehlungen nicht folgen, die S auf Kosten von R übervorteilen. Wir betrachten das Problem aus der Sicht von S und beschreiben Signalschemata, die S einen möglichst großen Nutzen garantieren.
Zuerst betrachten wir den Offline-Fall. Hier erfährt S den kompletten Zustand der Welt und schickt daraufhin ein Signal an R. Wir betrachten ein Szenario mit einer beschränkten Anzahl k ≤ n Signale. Mit nur k Signalen kann S höchstens k verschiedene Aktionen empfehlen. Für verschiedene symmetrische Instanzen beschreiben wir einen Polynomialzeitalgorithmus für die Berechnung eines optimalen Signalschemas mit k Signalen.
Weiterhin betrachten wir eine Teilmenge von Instanzen, in denen die Typen aus bekannten, unabhängigen Verteilungen gezogen werden. Wir beschreiben Polynomialzeitalgorithmen, die ein Signalschema mit k Signalen berechnen, das einen konstanten Approximationsfaktor im Verhältnis zum optimalen Signalschema mit k Signalen garantiert.
Im Online-Fall werden die Aktionstypen einzeln in Runden aufgedeckt. Nach Betrachtung der aktuellen Aktion sendet S ein Signal und R muss sofort durch Wahl oder Ablehnung der Aktion darauf reagieren. Der Prozess endet mit der Wahl einer Aktion. Andernfalls wird der nächste Aktionstyp aufgedeckt und vorherige Aktionen können nicht mehr gewählt werden. Als Richtwert für unsere Online-Signalschemata verwenden wir das beste Offline-Signalschema.
Zuerst betrachten wir ein Szenario mit unabhängigen Verteilungen. Wir zeigen, wie ein optimales Signalschema in Polynomialzeit bestimmt werden kann. Jedoch gibt es Beispiele, bei denen S – anders als im Offline-Fall – im Online-Fall keinen positiven Wert erzielen kann. Wir betrachten daraufhin eine Teilmenge der Instanzen, für die ein einfaches Signalschema einen konstanten Approximationsfaktor garantiert und zeigen dessen Optimalität.
Zusätzlich betrachten wir 16 verschiedene Szenarien mit unterschiedlichem Level an Information für S und R und unterschiedlichen Zielfunktionen für S und R unter der Annahme, dass die Aktionstypen a priori unbekannt sind, aber in uniform zufälliger Reihenfolge aufgedeckt werden. Für 14 Fälle beschreiben wir Signalschemata mit konstantem Approximationsfaktor. Solche Schemata existieren für die verbleibenden beiden Fälle nicht. Zusätzlich zeigen wir für die meistern Fälle, dass die beschriebenen Approximationsgarantien optimal sind.
Im zweiten Teil betrachten wir eine Online-Variante von Delegated Search. Hier besitzt nun R Commitment Power. Die Aktionstypen werden aus bekannten, unabhängigen Verteilungen gezogen. Bevor S die realisierten Typen beobachtet, legt R sich auf ein Akzeptanzschema φ fest. Für jeden Typen gibt φ an, mit welcher Wahrscheinlichkeit R diesen akzeptiert. Folglich versucht S, eine Aktion mit einem guten Typen für sich selbst zu finden, der von R akzeptiert wird. Da der Prozess online abläuft, muss S für jede Aktion einzeln entscheiden, diese vorzuschlagen oder zu verwerfen. Nur empfohlene Aktionen können von R ausgewählt werden.
Für den Offline-Fall sind für identisch verteilte Aktionstypen konstante Approximationsfaktoren im Vergleich zu einer Aktion mit optimalem Wert für R bekannt. Wir zeigen, dass R im Online-Fall im Allgemeinen nur eine Θ(1/n)-Approximation erzielen kann. Der Richtwert ist der erwartete Wert für eine eindimensionale Online-Suche von R.
Da für die Schranke eine exponentielle Diskrepanz in den Werten der Typen für S benötigt wird, betrachten wir parametrisierte Instanzen. Die Parameter beschränken die Werte für S bzw. das Verhältnis der Werte für R und S. Wir zeigen (beinahe) optimale logarithmische Approximationsfaktoren im Bezug auf diese Parameter, die von effizient berechenbaren Schemata garantiert werden.
The mobile games business is an ever-increasing sub-sector of the entertainment industry. Due to its high profitability but also high risk and competitive atmosphere, game publishers need to develop strategies that allow them to release new products at a high rate, but without compromising the already short lifespan of the firms' existing games. Successful game publishers must enlarge their user base by continually releasing new and entertaining games, while simultaneously motivating the current user base of existing games to remain active for more extended periods. Since the core-component reuse strategy has proven successful in other software products, this study investigates the advantages and drawbacks of this strategy in mobile games. Drawing on the widely accepted Product Life Cycle concept, the study investigates whether the introduction of a new mobile game built with core-components of an existing mobile game curtails the incumbent's product life cycle. Based on real and granular data on the gaming activity of a popular mobile game, the authors find that by promoting multi-homing (i.e., by smartly interlinking the incumbent and new product with each other so that users start consuming both games in parallel), the core-component reuse strategy can prolong the lifespan of the incumbent game.
Machine learning (ML) techniques have evolved rapidly in recent years and have shown impressive capabilities in feature extraction, pattern recognition, and causal inference. There has been an increasing attention to applying ML to medical applications, such as medical diagnosis, drug discovery, personalized medicine, and numerous other medical problems. ML-based methods have the advantage of processing vast amounts of data.
With an ever increasing amount of medical data collection and large, inter-subject variability in the medical data, automated data processing pipelines are very much desirable since it is laborious, expensive, and error-prone to rely solely on human processing. ML methods have the potential to uncover interesting patterns, unravel correlations between complex features, learn patient-specific representations, and make accurate predictions. Motivated by these promising aspects, in this thesis, I present studies where I have implemented deep neural networks for the early diagnosis of epilepsy based on electroencephalography (EEG) data and brain tumor detection based on magnetic resonance spectroscopy (MRS) data.
In the project for early diagnosis of epilepsy, we are dealing with one of the most common neurological disorders, epilepsy, which is characterized by recurrent unprovoked seizures. It can be triggered by a variety of initial brain injuries and manifests itself after a time window which is called the latent period. During this period, a cascade of structural and functional brain alterations takes place leading to an increased seizure susceptibility.
The development and extension of brain tissue capable of generating spontaneous seizures is defined as epileptogenesis (EPG).
Detecting the presence of EPG provides a precious opportunity for targeted early medical interventions and, thus, can slow down or even halt the disease progression. In order to study brain signals in this latent window, animal epilepsy models are used to provide valuable data as it is extremely difficult to obtain this data from human patients. The aim of this study is to discover biomarkers of EPG using animal models and then to find the equivalent and counterparts in human patients' data. However, the EEG features for EPG are not well-understood and there is not a sufficiently large amount of annotated data for ML-based algorithms. To approach this problem, firstly, I utilized the timestamp information of the recorded EEG from an animal epilepsy model where epilepsy is induced by an electrical stimulation. The timestamp serves as a form of weak supervision, i.e., before and after the stimulation. Secondly, I implemented a deep residual neural network and trained it with a binary classification task to distinguish the EEG signals from these two phases. After obtaining a high discriminative ability on the binary classification task, I proposed to divide further the time span after the stimulation for a three-class classification, aiming to detect possible stages of the progression of the latent EPG phase. I have shown that the model can distinguish EEG signals at different stages of EPG with high accuracy and generalization ability. I have also demonstrated that some of the learned features from the network are clinically relevant.
In the task of detecting brain tumors based on MRS data, I first proposed to apply a deep neural network on the MRS data collected from over 400 patients for a binary classification task. To combat the challenge of noisy labeling, I developed a distillation step to filter out relatively ``cleanly'' labeled samples. A mixing-based data augmentation method was also implemented to expand the size of the training set. All the experiments were designed to be conducted with a leave-patient-out scheme to ensure the generalization ability of the model. Averaged across all leave-patient-out cross-validation sets, the proposed method performed on par with human neuroradiologists, while outperforming other baseline methods. I have demonstrated the distillation effect on the MNIST data set with manually-introduced label noise as well as providing visualization of the input influences on the final classification through a class activation map method.
Moreover, I have proposed to aggregate information at the subject level, which could provide more information and insights. This is inspired by the concept of multiple instance learning, where instance-level labels are not required and which is more tolerant to noisy labeling. I have proposed to generate data bags consisting of instances from each patient and also proposed two modules to ensure permutation invariance, i.e., an attention module and a pooling module. I have compared the performance of the network in different cases, i.e., with and without permutation-invariant modules, with and without data augmentation, single-instance-based and multiple-instance-based learning and have shown that neural networks equipped with the proposed attention or pooling modules can outperform human experts.