Refine
Year of publication
- 2024 (117)
- 2023 (607)
- 2022 (1669)
- 2021 (1613)
- 2020 (1869)
- 2019 (2063)
- 2018 (2201)
- 2017 (1963)
- 2016 (1796)
- 2015 (1631)
- 2014 (1387)
- 2013 (1317)
- 2012 (1321)
- 2011 (1133)
- 2010 (705)
- 2009 (893)
- 2008 (459)
- 2007 (487)
- 2006 (840)
- 2005 (842)
- 2004 (844)
- 2003 (513)
- 2002 (421)
- 2001 (497)
- 2000 (557)
- 1999 (599)
- 1998 (562)
- 1997 (453)
- 1996 (386)
- 1995 (445)
- 1994 (253)
- 1993 (58)
- 1992 (49)
- 1991 (23)
- 1990 (4)
- 1989 (11)
- 1988 (7)
- 1987 (10)
- 1986 (9)
- 1985 (4)
- 1984 (7)
- 1983 (8)
- 1982 (2)
- 1981 (2)
- 1980 (1)
Document Type
- Article (21066)
- Doctoral Thesis (3156)
- Postprint (2347)
- Monograph/Edited Volume (1217)
- Other (657)
- Review (618)
- Preprint (529)
- Conference Proceeding (456)
- Part of a Book (243)
- Working Paper (179)
- Habilitation Thesis (75)
- Master's Thesis (61)
- Part of Periodical (45)
- Report (13)
- Bachelor Thesis (8)
- Journal/Publication series (6)
- Contribution to a Periodical (4)
- Moving Images (2)
- Lecture (1)
Language
- English (30683) (remove)
Keywords
- climate change (178)
- Germany (100)
- machine learning (80)
- diffusion (76)
- German (68)
- morphology (67)
- Arabidopsis thaliana (65)
- anomalous diffusion (58)
- stars: massive (56)
- Climate change (55)
Institute
- Institut für Physik und Astronomie (4955)
- Institut für Biochemie und Biologie (4683)
- Institut für Geowissenschaften (3307)
- Institut für Chemie (2909)
- Institut für Mathematik (1869)
- Department Psychologie (1480)
- Institut für Ernährungswissenschaft (1030)
- Department Linguistik (1006)
- Wirtschaftswissenschaften (858)
- Institut für Informatik und Computational Science (839)
Deep metric learning employs deep neural networks to embed instances into a metric space such that distances between instances of the same class are small and distances between instances from different classes are large. In most existing deep metric learning techniques, the embedding of an instance is given by a feature vector produced by a deep neural network and Euclidean distance or cosine similarity defines distances between these vectors. This paper studies deep distributional embeddings of sequences, where the embedding of a sequence is given by the distribution of learned deep features across the sequence. The motivation for this is to better capture statistical information about the distribution of patterns within the sequence in the embedding. When embeddings are distributions rather than vectors, measuring distances between embeddings involves comparing their respective distributions. The paper therefore proposes a distance metric based on Wasserstein distances between the distributions and a corresponding loss function for metric learning, which leads to a novel end-to-end trainable embedding model. We empirically observe that distributional embeddings outperform standard vector embeddings and that training with the proposed Wasserstein metric outperforms training with other distance functions.
A central insight from psychological studies on human eye movements is that eye movement patterns are highly individually characteristic. They can, therefore, be used as a biometric feature, that is, subjects can be identified based on their eye movements. This thesis introduces new machine learning methods to identify subjects based on their eye movements while viewing arbitrary content. The thesis focuses on probabilistic modeling of the problem, which has yielded the best results in the most recent literature. The thesis studies the problem in three phases by proposing a purely probabilistic, probabilistic deep learning, and probabilistic deep metric learning approach. In the first phase, the thesis studies models that rely on psychological concepts about eye movements. Recent literature illustrates that individual-specific distributions of gaze patterns can be used to accurately identify individuals. In these studies, models were based on a simple parametric family of distributions. Such simple parametric models can be robustly estimated from sparse data, but have limited flexibility to capture the differences between individuals. Therefore, this thesis proposes a semiparametric model of gaze patterns that is flexible yet robust for individual identification. These patterns can be understood as domain knowledge derived from psychological literature. Fixations and saccades are examples of simple gaze patterns. The proposed semiparametric densities are drawn under a Gaussian process prior centered at a simple parametric distribution. Thus, the model will stay close to the parametric class of densities if little data is available, but it can also deviate from this class if enough data is available, increasing the flexibility of the model. The proposed method is evaluated on a large-scale dataset, showing significant improvements over the state-of-the-art. Later, the thesis replaces the model based on gaze patterns derived from psychological concepts with a deep neural network that can learn more informative and complex patterns from raw eye movement data. As previous work has shown that the distribution of these patterns across a sequence is informative, a novel statistical aggregation layer called the quantile layer is introduced. It explicitly fits the distribution of deep patterns learned directly from the raw eye movement data. The proposed deep learning approach is end-to-end learnable, such that the deep model learns to extract informative, short local patterns while the quantile layer learns to approximate the distributions of these patterns. Quantile layers are a generic approach that can converge to standard pooling layers or have a more detailed description of the features being pooled, depending on the problem. The proposed model is evaluated in a large-scale study using the eye movements of subjects viewing arbitrary visual input. The model improves upon the standard pooling layers and other statistical aggregation layers proposed in the literature. It also improves upon the state-of-the-art eye movement biometrics by a wide margin. Finally, for the model to identify any subject — not just the set of subjects it is trained on — a metric learning approach is developed. Metric learning learns a distance function over instances. The metric learning model maps the instances into a metric space, where sequences of the same individual are close, and sequences of different individuals are further apart. This thesis introduces a deep metric learning approach with distributional embeddings. The approach represents sequences as a set of continuous distributions in a metric space; to achieve this, a new loss function based on Wasserstein distances is introduced. The proposed method is evaluated on multiple domains besides eye movement biometrics. This approach outperforms the state of the art in deep metric learning in several domains while also outperforming the state of the art in eye movement biometrics.
Different properties of programs, implemented in Constraint Handling Rules (CHR), have already been investigated. Proving these properties in CHR is fairly simpler than proving them in any type of imperative programming language, which triggered the proposal of a methodology to map imperative programs into equivalent CHR. The equivalence of both programs implies that if a property is satisfied for one, then it is satisfied for the other. The mapping methodology could be put to other beneficial uses. One such use is the automatic generation of global constraints, at an attempt to demonstrate the benefits of having a rule-based implementation for constraint solvers.
Say it with double flowers
(2020)
Every year, lovers world-wide rely on mutants to show their feelings on Valentine's Day. This is because many of the most popular ornamental flowering plants have been selected to form extra petals at the expense of reproductive organs to enhance their attractiveness and aesthetic value to humans. This so-called 'double flower' (DF) phenotype, first described more than 2000 years ago (Meyerowitz et al., 1989) is present, for example, in many modern roses, carnations, peonies, and camellias. Gattolin et al. (2020) now identify a unifying explanation for the molecular basis of many of these DF cultivars.
Phytochemical investigation of the dichloromethane/methanol (1:1) extract of the roots of Bulbine frutescens led to the isolation of a new xanthone, 8-hydroxy-6-methylxanthone-1-carboxylic acid (1) and a new phenylanthraquinone, 6',8-O-dimethylknipholone (2) along with six known compounds. The structures were elucidated on the basis of NMR and MS spectral data analyses. The structure of compound 1 was confirmed through X-ray crystallography which was then used as a reference to propose the revision of the structures of six seco-anthraquinones into xanthones. The isolated compounds were evaluated for cytotoxicity against human cervix carcinoma KB-3-1 cells with the phenylanthraquinone knipholone being the most active (IC50 = 0.43 mu M). Two semi-synthetic knipholone derivatives, knipholone Mannich base and knipholone-1,3-oxazine, were prepared and tested for cytotoxic activity; both showed moderate activities (IC50 value of 1.89 and 2.50 mu M, respectively). (C) 2014 Phytochemical Society of Europe. Published by Elsevier B.V. All rights reserved.
A new phenylanthrone, named knipholone cyclooxanthrone and a dimeric anthraquinone, 10-methoxy-10,7'-(chrysophanol anthrone)-chrysophanol were isolated from the roots of Kniphofia foliosa together with the rare naphthalene glycoside, dianellin. The structures were determined by NMR and mass spectroscopic techniques. The compounds showed antiplasmodial activities against the chloroquine-resistant (W2) and chloroquine-sensitive (D6) strains of Plasmodium falciparum with 10-methoxy-10,7'-(chrysophanol anthrone)-chrysophanol being the most active with IC50 values of 1.17 +/- 0.12 and 4.07 +/- 1.54 mu g/ml, respectively.
We report on the gamma-ray activity of the blazar Mrk 501 during the first 480 days of Fermi operation. We find that the average Large Area Telescope (LAT) gamma-ray spectrum of Mrk 501 can be well described by a single power-law function with a photon index of 1.78 +/- 0.03. While we observe relatively mild flux variations with the Fermi-LAT (within less than a factor of two), we detect remarkable spectral variability where the hardest observed spectral index within the LAT energy range is 1.52 +/- 0.14, and the softest one is 2.51 +/- 0.20. These unexpected spectral changes do not correlate with the measured flux variations above 0.3 GeV. In this paper, we also present the first results from the 4.5 month long multifrequency campaign (2009 March 15-August 1) on Mrk 501, which included the Very Long Baseline Array (VLBA), Swift, RXTE, MAGIC, and VERITAS, the F-GAMMA, GASP-WEBT, and other collaborations and instruments which provided excellent temporal and energy coverage of the source throughout the entire campaign. The extensive radio to TeV data set from this campaign provides us with the most detailed spectral energy distribution yet collected for this source during its relatively low activity. The average spectral energy distribution of Mrk 501 is well described by the standard one-zone synchrotron self-Compton (SSC) model. In the framework of this model, we find that the dominant emission region is characterized by a size less than or similar to 0.1 pc (comparable within a factor of few to the size of the partially resolved VLBA core at 15-43 GHz), and that the total jet power (similar or equal to 10(44) erg s(-1)) constitutes only a small fraction (similar to 10(-3)) of the Eddington luminosity. The energy distribution of the freshly accelerated radiating electrons required to fit the time-averaged data has a broken power-law form in the energy range 0.3 GeV-10 TeV, with spectral indices 2.2 and 2.7 below and above the break energy of 20 GeV. We argue that such a form is consistent with a scenario in which the bulk of the energy dissipation within the dominant emission zone of Mrk 501 is due to relativistic, proton-mediated shocks. We find that the ultrarelativistic electrons and mildly relativistic protons within the blazar zone, if comparable in number, are in approximate energy equipartition, with their energy dominating the jet magnetic field energy by about two orders of magnitude.
Observations of the young supernova remnant RX J1713.7-3946 with the fermi large area telescope
(2011)
We present observations of the young supernova remnant (SNR) RX J1713.7-3946 with the Fermi Large Area Telescope (LAT). We clearly detect a source positionally coincident with the SNR. The source is extended with a best-fit extension of 0 degrees.55 +/- 0 degrees.04 matching the size of the non-thermal X-ray and TeV gamma-ray emission from the remnant. The positional coincidence and the matching extended emission allow us to identify the LAT source with SNR RX J1713.7-3946. The spectrum of the source can be described by a very hard power law with a photon index of Gamma = 1.5 +/- 0.1 that coincides in normalization with the steeper H. E. S. S.-detected gamma-ray spectrum at higher energies. The broadband gamma-ray emission is consistent with a leptonic origin as the dominant mechanism for the gamma-ray emission.
This paper offers a new theoretical framework for studying the problem of generations and social change in contemporary Iran. It offers a model which is called „articulation of cultural modes“. The paper agrees with Ronald Inglehart that ‘culture’ is now playing a more dominant role in the social formation of current societies, as ‘technology’ once did in the modern era. But it goes one step further by arguing that culture cannot be approached as a holistic concept building on a comprehensive theoretical framework.
Combining the advection-diffusion equation approach with Monte Carlo simulations we study chaperone driven polymer translocation of a stiff polymer through a nanopore. We demonstrate that the probability density function of first passage times across the pore depends solely on the Peclet number, a dimensionless parameter comparing drift strength and diffusivity. Moreover it is shown that the characteristic exponent in the power-law dependence of the translocation time on the chain length, a function of the chaperone-polymer binding energy, the chaperone concentration, and the chain length, is also effectively determined by the Peclet number. We investigate the effect of the chaperone size on the translocation process. In particular, for large chaperone size, the translocation progress and the mean waiting time as function of the reaction coordinate exhibit pronounced sawtooth-shapes. The effects of a heterogeneous polymer sequence on the translocation dynamics is studied in terms of the translocation velocity, the probability distribution for the translocation progress, and the monomer waiting times. (C) 2011 American Institute of Physics.
Ionic guest in ionic host
(2022)
Ionosilica ionogels, i.e. composites consisting of an ionic liquid (IL) guest confined in an ionosilica host matrix, were synthesized via a non-hydrolytic sol-gel procedure from a tris-trialcoxysilylated amine precursor using the IL [BMIM]NTf2 as solvent. Various ionosilica ionogels were prepared starting from variable volumes of IL in the presence of formic acid. The resulting brittle and nearly colourless monoliths are composed of different amounts of IL guests confined in an ionosilica host as evidenced via thermogravimetric analysis, FT-IR, and C-13 CP-MAS solid-state NMR spectroscopy. In the following, we focused on confinement effects between the ionic host and guest. Special host-guest interactions between the IL guest and the ionosilica host were evidenced by H-1 solid-state NMR, Raman spectroscopy, and broadband dielectric spectroscopy (BDS) measurements. The three techniques indicate a strongly reduced ion mobility in the ionosilica ionogel composites containing small volume fractions of confined IL, compared to conventional silica-based ionogels. We conclude that the ionic ionosilica host stabilizes an IL layer on the host surface; this then results in a strongly reduced ion mobility compared to conventional silica hosts. The ion mobility progressively increases for systems containing higher volume fractions of IL and finally reaches the values observed in conventional silica based ionogels. These results therefore point towards strong interactions and confinement effects between the ionic host and the ionic guest on the ionosilica surface. Furthermore, this approach allows confining high volume fractions of IL into self-standing monoliths while preserving high ionic conductivity. These effects may be of interest in domains where IL phases must be anchored on solid supports to avoid leaching or IL spilling, e.g., in catalysis, in gas separation/sequestration devices or for the elaboration of solid electrolytes for (lithium-ion) batteries and supercapacitors.
One of the major challenges in chemical synthesis is to trigger and control a specific reaction route leading to a specific final product, while side products are avoided. Methodologies based on resonant processes at the molecular level, for example, photochemistry, offer the possibility of inducing selective reactions. Electrons at energies below the molecular ionization potential (<10 eV) are known to dissociate molecules via resonant processes with higher cross sections and specificity than photons. Here we show that even subexcitation electrons with energies as low as 1 eV produce ethylene and acetylene from dimethyl sulfide in competing reactions. However, the production of ethylene can specifically be targeted by controlling the energy of electrons (similar to 3 to 4 eV). Finally, pure ethylene is selectively desorbed by heating the substrate from 90 to 105 K. Beyond the synthesis of these versatile hydrocarbons for various industrial applications from a biogenic sulfur compound, our findings demonstrate the feasibility of electron induced selective chemistry applicable on the nanoscale.
The 11 July 1889 Chilik earthquake (M-w 8.0-8.3) forms part of a remarkable sequence of large earthquakes in the late nineteenth and early twentieth centuries in the northern Tien Shan. Despite its importance, the source of the 1889 earthquake remains unknown, though the macroseismic epicenter is sited in the Chilik valley, similar to 100 km southeast of Almaty, Kazakhstan (similar to 2 million population). Several short fault segments that have been inferred to have ruptured in 1889 are too short on their own to account for the estimated magnitude. In this paper we perform detailed surveying and trenching of the similar to 30 km long Saty fault, one of the previously inferred sources, and find that it was formed in a single earthquake within the last 700 years, involving surface slip of up to 10 m. The scarp-forming event, likely to be the 1889 earthquake, was the only surface-rupturing event for at least 5000 years and potentially for much longer. From satellite imagery we extend the mapped length of fresh scarps within the 1889 epicentral zone to a total of similar to 175 km, which we also suggest as candidate ruptures from the 1889 earthquake. The 175 km of rupture involves conjugate oblique left-lateral and right-lateral slip on three separate faults, with step overs of several kilometers between them. All three faults were essentially invisible in the Holocene geomorphology prior to the last slip. The recurrence interval between large earthquakes on any of these faults, and presumably on other faults of the Tien Shan, may be longer than the timescale over which the landscape is reset, providing a challenge for delineating sources of future hazard.
Background There is scant information on the breastmilk vitamin A (BMVA) concentration of lactating women in developing countries, partly due to lack of methods applicable in-field. Objective To assess BMVA concentrations of samples collected from lactating women of children aged 6-23 months, in Mecha district, Ethiopia. Subjects/methods Data on socio-demographic and anthropometric characteristics were collected from randomly selected lactating women (n = 104). Breast milk samples were collected and vitamin A concentrations were analyzed using HPLC and iCheck FLUORO then the two measurements were compared. Results The prevalence of underweight (BMI < 18.5 kg/m(2)) among lactating women was 17%. Seventy six percent of the BMVA values were < 1.05 mu mol/l and 81% were < 8 mu g/g fat. The mean BMVA concentration accounted to 41% of the estimated average value for mothers in developing countries. The BMVA values from HPLC and iCheck were correlated (r = 0.59, p = < 0.001), but it was not strong. Conclusions The result indicates the low vitamin A status of the lactating women and their children. It further indicates that intake assessments should not use average BMVA composition. The possibility of using iCheck for monitoring interventions designed to improve vitamin A status of lactating women with low BMVA requires further investigation.
We establish elements of a new approach to ellipticity and parametrices within operator algebras on manifolds with higher singularities, only based on some general axiomatic requirements on parameter-dependent operators in suitable scales of spaes. The idea is to model an iterative process with new generations of parameter-dependent operator theories, together with new scales of spaces that satisfy analogous requirements as the original ones, now on a corresponding higher level. The "full" calculus involves two separate theories, one near the tip of the corner and another one at the conical exit to infinity. However, concerning the conical exit to infinity, we establish here a new concrete calculus of edge-degenerate operators which can be iterated to higher singularities.
We establish elements of a new approch to ellipticity and parametrices within operator algebras on a manifold with higher singularities, only based on some general axiomatic requirements on parameter-dependent operators in suitable scales of spaces. The idea is to model an iterative process with new generations of parameter-dependent operator theories, together with new scales of spaces that satisfy analogous requirements as the original ones, now on a corresponding higher level. The “full” calculus is voluminous; so we content ourselves here with some typical aspects such as symbols in terms of order reducing families, classes of relevant examples, and operators near the conical exit to infinity.
Linked Open Data (LOD) comprises very many and often large public data sets and knowledge bases. Those datasets are mostly presented in the RDF triple structure of subject, predicate, and object, where each triple represents a statement or fact. Unfortunately, the heterogeneity of available open data requires significant integration steps before it can be used in applications. Meta information, such as ontological definitions and exact range definitions of predicates, are desirable and ideally provided by an ontology. However in the context of LOD, ontologies are often incomplete or simply not available. Thus, it is useful to automatically generate meta information, such as ontological dependencies, range definitions, and topical classifications. Association rule mining, which was originally applied for sales analysis on transactional databases, is a promising and novel technique to explore such data. We designed an adaptation of this technique for min-ing Rdf data and introduce the concept of “mining configurations”, which allows us to mine RDF data sets in various ways. Different configurations enable us to identify schema and value dependencies that in combination result in interesting use cases. To this end, we present rule-based approaches for auto-completion, data enrichment, ontology improvement, and query relaxation. Auto-completion remedies the problem of inconsistent ontology usage, providing an editing user with a sorted list of commonly used predicates. A combination of different configurations step extends this approach to create completely new facts for a knowledge base. We present two approaches for fact generation, a user-based approach where a user selects the entity to be amended with new facts and a data-driven approach where an algorithm discovers entities that have to be amended with missing facts. As knowledge bases constantly grow and evolve, another approach to improve the usage of RDF data is to improve existing ontologies. Here, we present an association rule based approach to reconcile ontology and data. Interlacing different mining configurations, we infer an algorithm to discover synonymously used predicates. Those predicates can be used to expand query results and to support users during query formulation. We provide a wide range of experiments on real world datasets for each use case. The experiments and evaluations show the added value of association rule mining for the integration and usability of RDF data and confirm the appropriateness of our mining configuration methodology.