Publication Server of Hochschule Darmstadt
Not a member yet
259 research outputs found
Sort by
Applying machine intelligence in practice
Full text linkThe relevance of Machine Intelligence, a.k.a. Artificial Intelligence (AI), is undisputed at the present time. This is not only due to AI successes in research but, more prominently, its use in day-to-day practice. In 2014, we started a series of annual workshops at the Leibniz Zentrum für Informatik, Schloss Dagstuhl, Germany, initially focussing on Corporate Semantic Web and later widening the scope to Applied Machine Intelligence. This article presents a number of AI applications from various application domains, including medicine, industrial manufacturing and the insurance sector. Best practices, current trends, possibilities and limitations of new AI approaches for developing AI applications are also presented. Focus is put on the areas of natural language processing, ontologies and machine learning. The article concludes with a summary and outlook
Investigations of Free Space Optical Communications Under Real-World Atmospheric Conditions
Full text linkDue to the increasing demand for higher bandwidth in modern communication systems, conventional networks are continuously expanded with new technologies to improve coverage. Free space optical communications (FSOC) shows some significant advantages concerning system setup time in comparison with the classical fiber optical systems on one hand, substantial spectral bandwidth and performances in comparison with the wireless systems under certain conditions on the other hand. This makes this technology not only a reasonable extension for metropolitan area networks but also provides the capability to set up a network after an outage in case of natural disaster quickly. But transmitting data by using FSOC involves some limiting factors that have to be considered prior to each installation. Since the atmospheric channel is not static, the influence of changing weather conditions or industrial smog have a significant impact on the available bitrate. A simulation platform is developed and presented in this paper for investigation of FSOC considering these circumstances. Regarding the atmospheric channel, turbulence, distance-dependent beam divergence, and applied modulation schemes, a general overview of the capabilities is presented and discussed. The insight of this paper should help to make a decision under which preconditions either the FSOC provides a meaningful application possibility, or the limiting factors become too crucial and other technologies must be considered
Controlling the false discovery exceedance for heterogeneous tests
Full text linkSeveral classical methods exist for controlling the false discovery exceedance (FDX) for large-scale multiple testing problems, among them the Lehmann-Romano procedure (Lehmann and Romano 2005) ([LR] below) and the Guo-Romano procedure (Guo and Romano 2007) ([GR] below). While these two procedures are the most prominent, they were
originally designed for homogeneous test statistics, that is, when the null distribution functions of the p-values Fi, 1 ≤ i ≤ m, are all equal. In many applications, however, the data are heterogeneous which leads to heterogeneous null distribution functions. Ignoring this heterogeneity induces a lack of power. In this paper, we develop three new procedures that incorporate the Fi’s, while maintaining rigorous FDX control. The heterogeneous version of [LR], denoted [HLR], is based on the arithmetic average of the Fi’s, while the heterogeneous version of [GR], denoted [HGR], is based on the geometric average of the Fi’s. We also introduce a procedure [PB], that is based on the Poisson-binomial distribution and that uniformly improves [HLR] and [HGR], at the price of a higher computational complexity. Perhaps surprisingly, this shows that, contrary to the known theory of false discovery rate (FDR) control under heterogeneity, the way to incorporate the Fi’s can be particularly simple in the case of FDX control, and does not require any further correction term. The performances of the new proposed procedures are illustrated by real and simulated data in two important heterogeneous settings: first, when the test statistics are continuous but
the p-values are weighted by some known independent weight vector, e.g., coming from co-data sets; second, when the test statistics are discretely distributed, as is the case for data representing frequencies or counts. Our new procedures are implemented in the R package FDX, see Junge and Döhler (2020)
Lessons Learned From Applications of the Stage Model of Self-Regulated Behavioral Change: A Review
Full text linkStage models are becoming increasingly popular in explaining change from current behavior to more environmentally friendly alternatives. We review empirical applications of a recently introduced model, the stage model of self-regulated behavioral change (SSBC). In the SSBC, change toward pro-environmental behavior takes place in four, qualitatively different stages (predecisional, preactional, actional, and postactional) which are each influenced by constructs taken from theories previously established to describe and predict pro-environmental behavior. We performed a systematic literature search to retrieve peer-reviewed SSBC-based studies. The review includes 10 studies published between 2013 and 2018, six of which employed a cross-sectional, three an interventional and one a correlational longitudinal design. The cross-sectional and longitudinal studies generally support the model, although there are some irregularities that warrant further investigation. The interventional studies found stage-tailored informational measures to be more effective than non-stage-tailored measures in promoting stage progression and behavioral change. Furthermore, we identified several challenges that researchers may face when applying the SSBC. These include whether and how to analyze multiple behavioral alternatives; how to address the challenge of measuring a comprehensive model while keeping questionnaire length manageable; selecting and defining the role of model constructs in a behavioral context while keeping results comparable; and establishing a validated and reliable tool to diagnose a person’s stage of change. Based on these insights, we develop recommendations for researchers designing SSBC studies, in order to support a founded and efficient advancement of the theory which will then serve both researchers and practitioners aiming to promote pro-environmental behavior
Self-lead tension release and evaluation of university education
Full text linkRelationships between physical self-leadership by using strategies of tension release and evaluations of university education were analysed. 312 students participated in the study and completed a questionnaire measuring two types of self-lead physical relaxation, i.e. common and advanced strategies of tension release, felt stress, satisfaction with studies, and personal commitment. Results show that common strategies of tension release were more frequently used than advanced strategies. Students who made frequent use of common strategies experienced less stress, more satisfaction and a higher degree of personal commitment than students who made little use of common strategies. The students’ satisfaction with studies and personal commitment did not differ depending on how frequently they used advanced strategies of tension release. However, students who frequently applied advanced strategies tended to feel more stress than students who made little use of advances strategies. Implications of these results are discussed
Automatische Erkennung von politischen Trends mit Twitter – brauchen wir Meinungsumfragen noch?
Full text linkMeinungsforschungsinstitute betreiben einen beträchtlichen Aufwand, um die Meinungstrends der Bevölkerung bezogen auf Politiker mit Telefon- und Straßenumfragen zu erfassen. Mit einer Studierendengruppe haben wir uns im Winter 2015/16 die Frage gestellt, ob es möglich ist, diesen Prozess zu automatisieren. Die Idee dahinter ist, dass die Plattform Twitter vielfach für politische Diskussionen genutzt wird. Da sich Tweets auf einen Umfang von 140 Zeichen beschränken und das jeweilige Thema durch Hashtags meist eindeutig zugeordnet werden kann, scheinen sich Twitter-Daten gut für eine automatische Sentiment-Analyse zu eignen. Mit Sentiment-Analyse-Methoden kann man diese Tweets automatisch in positive und negative Meinungsäußerungen klassifizieren. Wir haben dafür einen Twitter-Crawler und Sentiment-Analyse in der Programmiersprache Python implementiert. Anschließend haben wir über einen Zeitraum von vier Wochen Tweets zu Politikern gesammelt und die Ergebnisse der Meinungsanalysen visualisiert. Schließlich haben wir unsere Ergebnisse mit dem ZDF-Politbarometer verglichen