Refine
Year of publication
- 2017 (65) (remove)
Document type
- Article (peer-reviewed) (27)
- Bachelor Thesis (16)
- Academic Papers (9)
- Master's Thesis (5)
- Report (5)
- Working Paper (2)
- Conference Proceeding (1)
Has full text
- Yes (65) (remove)
Keywords
- Electrical impedance tomography (4)
- HSV colour space (2)
- Maschinelles Lernen (2)
- Medical imaging (2)
- Smoke detection (2)
- 1,2,4-triazinone (1)
- 16S rRNA gene sequencing (1)
- 4-bromo-2-chlorophenyl (1)
- ACCESS (1)
- ARDS (1)
Course of studies
- IBW - Internationale Betriebswirtschaft (14)
- IMM - International Management (6)
- IBM - International Business Management (4)
- OMB - OnlineMedien (3)
- BPT - Bio- und Prozess-Technologie (1)
- EMBA - Executive Master of International Business Management (1)
- INM - Informatik (1)
- MEB - Medical Engineering (1)
- MZ - Mikromedizin (1)
- TMB - Wirtschaftsingenieurwesen - Technikmanagement (1)
Vergleich maschineller Lernalgorithmen in Anwendung auf die 3D-Gestenerkennung per Leap Motion
(2017)
Die Gestensteuerung ermöglicht eine intuitive sowie natürliche Interaktion zwischen Mensch und Maschine. Um die Bewegungen des Nutzers im dreidimensionalen Raum erkennen und klassifizieren zu können, sind Sensoren nötig, deren Signale es gilt zu verarbeiten. Diese Verarbeitung geschieht mittlerweile überwiegend durch Algorithmen des maschinellen Lernens, welche mit Hilfe von eingelernten Daten selbstständig Vorhersagen über neue Eingabesignale treffen können.
In dieser Arbeit werden die Eigenschaften und die damit zusammenhängende Erkennungsgenauigkeit solcher Algorithmen untersucht und verglichen. Dabei sollen die Support Vector Machine und das neuronale Netz berücksichtigt werden, welche mit Hilfe der Open Source Bibliotheken TensorFlow und scikit-learn implementiert wurden. Als Untersuchungsgegenstand dienen die Positionskoordinaten von insgesamt 350 dynamischen Gesten. Diese wurden mit dem Tiefenensor Leap Motion mittels einer eigens entwickelten Software aufgezeichnet und entsprechend vorverarbeitet. Auf diese Software, sowie den Prozess der Datensammlung, der Vorverarbeitung und letztendlich der Klassifizierung wird in der Arbeit ausführlicher eingegangen.
Die durchgeführten Tests zeigen, dass beide Algorithmen in der Lage waren eine Erkennungsgenauigkeit von bis zu 93,40% zu erreichen. Die Support Vector Machine schnitt durchschnittlich mit 87,15%, gegenüber dem neuronalen Netz mit 80,47%, besser ab.
Im Rahmen dieser Bachelorarbeit wird untersucht, ob kulturelle Faktoren einen Einfluss auf das Verhältnis zwischen in Deutschland lebenden türkischstämmigen Anwälten und Mandanten haben. Hierzu wird die ethnische Identität von Deutsch-Türken im Zusammenhang mit Generationsunterschieden untersucht. Zudem werden Akkulturationsmodelle, Kommunikationstheorien sowie Ziele und Möglichkeiten des ethnischen Marketings definiert.
Durch Anwendung qualitativer und quantitativer Forschungsmethoden werden jeweils aus Anwalts- und Mandantensicht die ethnische Identität sowie die für die Präferenz eines türkischstämmigen Anwalts relevanten Faktoren einzelner Einwanderergenerationen ermittelt.
Die Ergebnisse zeigen, dass die Anwalts-Mandanten-Beziehung stark von kulturellen Aspekten beeinflusst wird und dass Änderungen im anwaltlichen Ethno-Marketing notwendig sind, da diese nicht von türkischstämmigen Mandanten wahrgenommen werden.
The Revive Turbo Cleaner is a cleaning fluid which removes carbon deposits from inside petrol or diesel engine systems. It is distributed internationally through wholesalers and sales agents to workshops and private car owners. After the acquisition of Revive Automotive Solutions Ltd by Bartec Auto ID Ltd in summer 2016, the new owners decided to further develop the brand and to move into the promising German automotive aftermarket, as the Cleaner had shown good sales figures in the home market UK and in a number of international markets. This paper examines environmental factors influencing the German marketplace, analyses Revive’s actual customer segments including consumers and business clients, provides profiles of its competitors and substituting technologies, regards the characteristics of the specific industry and defines the product’s qualities in order to draw a conclusion on the appropriate Marketing Mix to employ for the German market. In advance to the analytical work, the paper critically considers the range of theoretical frameworks used for the analysis, namely PESTEL, Porter’s Five Forces Analysis, the SWOT Analysis and the Marketing Mix in order to define and update original concepts as wells as adapt them to the research requirements.
This thesis investigates the effect choice options in e-commerce applications have on consumers’ decision making. Previous research showed that a large number of options
can affect consumers negatively. However, the conditions for such choice overload are unclear. After reviewing the existing research, the amount of information (entropy)
contained in a choice set and individual differences were determined as possible influencing factors in an online environment. In a choice experiment, choice sets with
varying information loads and an assessment of the Big Five personality traits were used to test the impact of the two identified factors on choice avoidance behavior. Results from chi-square-tests and a logistic regression model suggest choice overload but without entropy having an effect. A logistic regression model revealed that extraverted consumers are easier overloaded. A low Neuroticism score was found to be related to less occurrence of a too-much-choice-effect. Consumers with a high Openness score on the other hand choose one of the presented options more often and were therefore less often
overwhelmed by the assortment. An interaction effect between personality and the amount of entropy was not found. These findings extend the research on choice overload and offer valuable input for marketers targeting consumers online.
Die Entwicklung zu Industrie 4.0 stellt neue Herausforderungen an alle Bereiche eines Unternehmens. Gerade dem Supply Chain Management werden neue Anforderungen gestellt, um bei der globalen Vernetzung der Supply Chains nicht den Anschluss zu verlieren. Die Vernetzung zu cyber-physischen Systemen erzeugt große Menge an Daten, Big Data, aus denen die richtigen Informationen gewonnen werden sollten. Industrie 4.0 wird auch Einfluss auf den Arbeitsalltag, die Organisation und die Mitarbeiter nehmen. Den enormen Potenzialen stehen Risiken und hohe Investitionen entgegen. Vor allem das Thema Datensicherheit spielt gerade in Deutschland, eine wichtige Rolle. Neue Technologien, wie die Blockchain, stehen noch am Anfang der Entwicklung, werden aber schnell Einfluss auf das industrielle Umfeld und somit auch auf das Supply Chain Management nehmen.
Mobile diagnostics – or mobile health in general – is highly appealing, not only for clinicians, but also for patients. It implies empowerment, in particular of those who are really in need, such as inhabitants of less developed regions within the world who have limited access to healthcare. It also implies simplification: Easy data management – a continuous flow of information. Therefore, development of miniaturized and highly integrated diagnostic systems allowing near patient “instant” diagnostics gain a lot of momentum since more than a decade. However, system integration requires time and a significant amount of investment. In addition, there is strong competition on resources from other emergent technologies, such as next generation sequencing which made the collection of e.g. human genome data less expensive and much faster. A more severe challenge is that mobile diagnostics require a change in healthcare management, e.g. towards integrated practice units. This, in turn, requires implementation of adequate reimbursement, standards of interoperability, training of staff, quality control. In 2010, Germany’s Federal Ministry of Education and Research (BMBF) launched the grant initiative Mobile Diagnostic Systems (MD, 2011─2015) as part of its high-tech strategy. MD aimed at generating knowledge on how microsystem technologies fit into German healthcare environments. On the basis of interviews with multidisciplinary MD actors, this thesis evaluated retrospectively how the publicly funded innovation network managed to overcome pre-defined external barriers of diffusion, including technology, regulatory affairs and market access. Retrospectives reveal internal barriers involving knowledge and technology transfer, negatively influencing generation of innovation. In particular, financing still represents a high hurdle for biotech innovators in Germany: Larger firms look predominately for market-ready or in-market technologies rather than prototypes and venture capitalists are rare or extremely risk-averse. Another important finding was, that actors involved were highly focused on individual work packages. This risks of not seeing the whole environment embedding MD. Consequently, potential opportunities may be missed, e.g. synergies with relatively close (DIALOC) or more distant initiatives (Global Health Delivery Project-based discussion rounds). This could be partly due to the fact that publicly funded networking activities provide less freedom-to-operate because of pre-defined milestones. In addition, further development of actors with respect to role playing (e.g. boundary spanning or innovation selling) is often not included in such “innovation packages”, but can help to maneuver change. Internal barriers need to be addressed first before targeting the major remaining external hurdle: Reimbursement. Although the latter was covered within MD, standardization of technology evaluation is still an unmet need which strongly influences the willingness-to-implement novel mobile diagnostics. Thus, the value added is to be demonstrated to justify adequate reimbursement. Achieving this goal can be successful, when innovation networking finds its path towards a common vision, e.g. towards value-based integrated healthcare. Pathfinding and visioning can be facilitated by process promoter with excellent network management capabilities. In addition, such a promoter could help to further develop engagement, openness and commitment of collaborators. Therefore, transfer of MD activities to established “top” networks or clusters is recommended for securing valuable knowledge generated. In this environment, an important next step – globalization of MD for ensuring future return on investment – could be triggered as well.
Since MD innovation was found to involve both product and service innovation, maneuvering change is particularly challenging for small and medium sized enterprises. These could benefit from engagement in innovation networking. Findings of this case study can help all direct and indirect actors in the field of MD innovation or in other high complex environments to reconsider pathfinding as well as role playing in networking.