Gutenberg Open Science (Univ. Mainz)
Not a member yet
11784 research outputs found
Sort by
Patientenmanagement bei chronischer Urtikaria und Komorbidität
No full text98 Seiten ; Illustrationen, Diagramm
On the loss landscape of deep neural networks
No full textAs a non-convex optimization problem, the training of deep neural networks remains poorly understood, and its success critically depends on the exact network architecture used. While the amount of new network architectures proposed in the last decade is staggering, only a handful of common patterns emerged that are shared by most successful architectures. First, using the smoothness of the optimization landscape as a heuristic for trainability, we investigate what network components render training difficult and how these patterns help alleviate such difficulties. We find that while giving networks their expressivity, deep stacks of nonlinear layers significantly increase the roughness of the optimization landscape as network depth increases. Developing prior work, we quantify this effect and show that for networks at initialization, the strength of this effect depends on the smoothness of the nonlinear layer used. We then demonstrate how residual connections and multi-path architectures reduce high frequencies in the optimization landscape, resulting in increased trainability. Second, we found that normalization layers combined with an adequate warm-up scheme compensate for the increasing roughness in lower layers by dynamically re-scaling the layer-wise gradients. We prove that in a properly normalized network, all layer-wise effective learning speeds align over time, compensating for even exponentially exploding gradients at initialization. Finally, we conduct an empirical study to determine the necessary nonlinear depth of a network to generalize effectively on common deep learning tasks. Surprisingly, we find that a shallow network extracted after training significantly outperforms a comparably shallow network trained from scratch, although their expressivity is exactly the same. We also observe that ensembles of both shallow and deep paths outperform comparable networks comprised of only deep paths, even when extracted after training. Using these insights, we aim to gain a deeper understanding of how to design deep neural networks with high trainability and strong generalization properties.x, 114, 2 Seiten ; Illustrationen, Diagramm
Mixed Methods Analyse des Handlungsfeldes der Physiotherapie bei erwachsenen Patienten mit palliativem Versorgungsbedarf auf der Intensivstation
No full textV, 128 Seiten ; Diagramm
Synthesis of carbohydrate antigens against serotype 3 streptococcus pneumoniae and supramolecular hydrogels crosslinked via cucurbit[8]uril host-guest complexation
No full text216 Seiten ; Illustrationen, Diagramm
Effects of chiral polypeptides on skyrmion stability and dynamics
No full textMagnetic skyrmions, topologically stabilized chiral spin textures in magnetic thin films, have garnered considerable interest due to their efficient manipulation and resulting potential as efficient nanoscale information carriers. One intriguing approach to address the challenge of tuning skyrmion properties involves using chiral molecules. Chiral molecules can locally manipulate magnetic properties by inducing magnetization through spin exchange interactions and by creating spin currents. Here, Magneto-Optical Kerr Effect (MOKE) microscopy is used to image the impact of chiral polypeptides on chiral magnetic structures. The chiral polypeptides shift the spin reorientation transition temperature, reduce thermal skyrmion motion, and alter the coercive field locally, enhancing skyrmion stability and thus enabling local control. These findings demonstrate the potential of chiral molecules to address challenges for skyrmion based devices, thus paving the way to applications such as the racetrack memory, reservoir computing and others
Raw data for "Mechanistic insights: correspondence on “Tuning co-operative energy transfer in copper(I) complexes using two-photon absorbing diimine-based ligand sensitizers”"
No full textIn a recent communication, Collins and co-workers presented a Cu(I) complex with photocatalytic activity under red light LED conditions, mainly for singlet oxygen-driven reactions. Guided by steady-state emission measurements with 800 nm excitation, the authors suggested that the underlying mechanism for the generation of the photoexcited key species is a simultaneous two-photon absorption via a virtual state. However, such a mechanism requires pulsed laser excitation and cannot compete when a conventional one-photon excitation is also feasible with the selected excitation wavelength range. Using several spectroscopic techniques and reactivity assays under different light color and intensity conditions, we unambiguously demonstrate that a conventional one-photon excitation followed by rather inefficient singlet oxygen generation (quantum yield < 5%) is responsible for the observed photoreactivity of the Cu(I) complex. In addition, we briefly summarize general mechanistic considerations, estimate typical photon densities required for a variety of two-photon mechanisms, highlight the importance of optical filters and impurities to avoid artifacts in the emission spectra, and present some guidelines for the differentiation between one- and two-photon mechanisms.Experiment
Repeated fractionation and umbel receptacle elongation explain the apparent “panicle with simple umbels” in Ferula species (Apiaceae)
Full text linkIntroduction: The carrot family (Apiaceae) is characterized by umbels with umbellets. Traditionally, these umbels are interpreted as inflorescences. Ontogenetic studies, however, indicate that they do not originate from inflorescence meristems but from flower-like floral unit meristems. These meristems repeatedly fractionate sub-meristems, which give rise to umbellets and flowers. Ferula species usually form double racemes with umbels with umbellets. Few species of the genus, previously grouped in the genus Dorema, however, present “panicles with simple umbels”.
Methods: Aiming to identify the developmental processes resulting in the different inflorescence appearance, we investigate inflorescence development in Ferula hezarlalehzarica (double racemes with umbels with umbellets) and Ferula aucheri (panicles with simple umbels).
Results: Both species are andromonoecious (perfect and staminate flowers) and produce huge yellow inflorescences. SEM studies confirm that they share the same developmental patterns. Their development starts with an inflorescence meristem segregating umbel meristems. These pass through two steps of fractionation generating first umbellet meristems and then flower meristems. F. aucheri differs from F. hezarlalehzarica by i) producing several lateral inflorescences apart from one terminal one and ii) extremely elongating the umbel receptacles, thereby separating the umbellets from each other. The unusual branches with simple umbels thus prove to be homologous to umbels with umbellets. Furthermore, F. aucheri shows some intermediate inflorescences with umbellets intermixed with umbels. Considering that umbels and umbellets only differ in one step of fractionation, we interpret this mixture as developmental lability.
Discussion: The study shows that meristem conditions define the character of the umbels as floral units and that developmental processes like fractionation, expansion, and elongation shape their outer appearance. It illustrates that inflorescences can be easily misinterpreted if only adult branching systems are investigated
Visual political communication on Instagram : a comparative study of Brazilian presidential elections
Full text linkIn today’s digital age, images have become powerful tools for politicians to engage with their voters on social media platforms. Visual content possesses a unique emotional appeal that often leads to increased user engagement. However, research on visual communication remains relatively limited, particularly in the Global South. This study aims to bridge this gap by employing a combination of computational methods and qualitative approach to investigate the visual communication strategies employed in a dataset of 11,263 Instagram posts by 19 Brazilian presidential candidates in 2018 and 2022 national elections. Through two studies, we observed consistent patterns across these candidates on their use of visual political communication. Notably, we identify a prevalence of celebratory and positively toned images. They also exhibit a strong sense of personalization, portraying candidates connected with their voters on a more emotional level. Our research also uncovers unique contextual nuances specific to the Brazilian political landscape. We note a substantial presence of screenshots from news websites and other social media platforms. Furthermore, text-edited images with portrayals emerge as a prominent feature. In light of these results, we engage in a discussion regarding the implications for the broader field of visual political communication. This article contributes by showing the ways Instagram was used in the digital political strategy of two fiercely polarized Brazilian elections, shedding light on the ever-evolving dynamics of visual political communication in the digital age. Finally, we propose avenues for future research in the field of political communication
