Johannes Kepler University of Linz

About: Johannes Kepler University of Linz is a education organization based out in Linz, Oberösterreich, Austria. It is known for research contribution in the topics: Computer science & Thin film. The organization has 6605 authors who have published 19243 publications receiving 385667 citations.

Self-Normalizing Neural Networks

Abstract: Deep Learning has revolutionized vision via convolutional neural networks (CNNs) and natural language processing via recurrent neural networks (RNNs). However, success stories of Deep Learning with standard feed-forward neural networks (FNNs) are rare. FNNs that perform well are typically shallow and, therefore cannot exploit many levels of abstract representations. We introduce self-normalizing neural networks (SNNs) to enable high-level abstract representations. While batch normalization requires explicit normalization, neuron activations of SNNs automatically converge towards zero mean and unit variance. The activation function of SNNs are "scaled exponential linear units" (SELUs), which induce self-normalizing properties. Using the Banach fixed-point theorem, we prove that activations close to zero mean and unit variance that are propagated through many network layers will converge towards zero mean and unit variance -- even under the presence of noise and perturbations. This convergence property of SNNs allows to (1) train deep networks with many layers, (2) employ strong regularization, and (3) to make learning highly robust. Furthermore, for activations not close to unit variance, we prove an upper and lower bound on the variance, thus, vanishing and exploding gradients are impossible. We compared SNNs on (a) 121 tasks from the UCI machine learning repository, on (b) drug discovery benchmarks, and on (c) astronomy tasks with standard FNNs and other machine learning methods such as random forests and support vector machines. SNNs significantly outperformed all competing FNN methods at 121 UCI tasks, outperformed all competing methods at the Tox21 dataset, and set a new record at an astronomy data set. The winning SNN architectures are often very deep. Implementations are available at: this http URL.

Ultrathin and lightweight organic solar cells with high flexibility

Abstract: Organic solar cells are promising for technological applications, as they are lightweight and mechanically robust. This study presents flexible organic solar cells that are less than 2 μm thick, have very low specific weight and maintain their photovoltaic performance under repeated mechanical deformation.

Morphology of polymer/fullerene bulk heterojunction solar cells

Abstract: Within the different organic photovoltaic devices the conjugated polymer/fullerene bulk heterojunction approach is one of the foci of today's research interest. These devices are highly dependent on the solid state nanoscale morphology of the two components (donor/acceptor) in the photoactive layer. The need for finely phase separated polymer–fullerene blends is expressed by the limited exciton diffusion length present in organic semiconductors. Typical distances that these photo-excitations can travel within a pristine material are around 10–20 nm. In an efficient bulk heterojunction the scale of phase separation is therefore closely related to the respective exciton diffusion lengths of the two materials involved. Once the excitons reach the donor/acceptor interface, the photoinduced charge transfer results in the charge separation. After the charges have been separated they require percolated pathways to the respective charge extracting electrodes in order to supply an external direct current. Thus also an effective charge transport relies on the development of a suitable nanomorphology i.e. bicontinuous interpenetrating phase structures within these blend films. The present feature article combines and summarizes the experimental findings on this nanomorphology–efficiency relationship.

Analysis of shared heritability in common disorders of the brain

Abstract: Disorders of the brain can exhibit considerable epidemiological comorbidity and often share symptoms, provoking debate about their etiologic overlap. We quantified the genetic sharing of 25 brain disorders from genome-wide association studies of 265,218 patients and 784,643 control participants and assessed their relationship to 17 phenotypes from 1,191,588 individuals. Psychiatric disorders share common variant risk, whereas neurological disorders appear more distinct from one another and from the psychiatric disorders. We also identified significant sharing between disorders and a number of brain phenotypes, including cognitive measures. Further, we conducted simulations to explore how statistical power, diagnostic misclassification, and phenotypic heterogeneity affect genetic correlations. These results highlight the importance of common genetic variation as a risk factor for brain disorders and the value of heritability-based methods in understanding their etiology.

Culture specific and cross-culturally generalizable implicit leadership theories: Are attributes of charismatic/transformational leadership universally endorsed?

Abstract: This study focuses on culturally endorsed implicit theories of leadership (CLTs). Although cross-cultural research emphasizes that different cultural groups likely have different conceptions of what leadership should entail, a controversial position is argued here: namely that attributes associated with charismatic/transformational leadership will be universally endorsed as contributing to outstanding leadership. This hypothesis was tested in 62 cultures as part of the Global Leadership and Organizational Behavior Effectiveness (GLOBE) Research Program. Universally endorsed leader attributes, as well as attributes that are universally seen as impediments to outstanding leadership and culturally contingent attributes are presented here. The results support the hypothesis that specific aspects of charismatic/transformational leadership are strongly and universally endorsed across cultures.