Zur Startseite von DrArbeit.de,
der deutschlandweiten Stellenbörse für Diplomarbeiten und Doktorarbeiten


Lupe2

Archiv - Stellenangebot

 

Dies ist ein Angebot aus der Datenbank von DrArbeit.de

Um die Datenbank komfortabel nach weiteren Angeboten durchsuchen zu können, klicken Sie einfach oben oder hier.

 
Archiv-Übersicht     Angebot Nr. 13584

Angebotsdatum: 1. September 2019
Art der Stelle: Doktorarbeit
Fachgebiet: Informatik > Medizininformatik
Titel des Themas: Visualization of Knowledge Graphs for Validation, Modification, and Optimization

Institut: Aarhus University, Department of Computer Science
Adresse:
Prof. Hans-Jörg Schulz
Åbogade 34
8200 Aarhus, Dänemark
Tel.:    Fax.:
Bundesland:
Homepage: http://phd.scitech.au.dk/for-applicants/apply-here/november-2019/visualization-of-knowledge-graphs-for-validation-modification-and-optimization/
E-Mail Kontakt: mail

Beschreibung: Applications are invited for a PhD fellowship/scholarship at Graduate School of Science and Technology, Aarhus University, Denmark, within the Computer Science program. The position is available from 1 February 2020 or later.

Research area and project description:
Machine learning methods generate models from training datasets. In many cases, these models are obscure and incomprehensible to the human user – who, as a result, remains unsure about what the models encode and whether to trust their predictions. While some uncertainty about their inner functioning may be acceptable when telling dog pictures from cat pictures, in other applications like medicine, autonomous driving, or air traffic control, it is not. For these applications, visualization holds the promise to play an important role in making machine learning results more transparent, traceable, and predictable.
This PhD project sets out to research novel visualization methods for a particular type of machine learning result – so-called Knowledge Graphs. Within an interdisciplinary research project, these graph representations are generated to capture workflows and their relevant context information in hospitals. It will be your task to first and foremost develop visual-interactive techniques to communicate these knowledge graphs to medical experts, so that they can validate and modify these graphs if necessary. Furthermore, you are to explore the possibility of using the resulting visualization as a means for monitoring and optimizing healthcare workflows in hospitals.
Research questions to pursue in this project are:

How to show the learned knowledge graph in a way that aligns with the mental map medical doctors have about their hospital’s workflows? In which ways do user preferences and professional background, as well as the task at hand influence this display?

What are good starting points for manual modifications of the learned knowledge graph? (e.g., circular or contradicting procedures within the same workflow) How to support manual editing of knowledge graphs by showing its implications in a What-If manner?

How to discern between fixed, generally accepted industry-wide procedures and the deviations and more nuanced approaches as employed in a particular hospital, in a particular ward, or only by a particular doctor? How can such a layered approach help to give feedback either to improve the model, or to improve a ward’s or a doctor’s practices?

Work on this PhD topic will be conducted as part of the Hospital@Night project – a grand solution project funded by the Innovationfund Denmark. Within this project, you will collaborate closely with machine learning specialists from the Data-Intensive Systems Group, healthcare IT experts from Systematic, and medical specialists from Aarhus University Hospital and Aalborg University Hospital.

Qualifications and specific competences:
To apply for the position, you must have a relevant Master’s degree and excellent computer programming skills. Prior experience in at least one of the following areas is of advantage: data visualization, data science, computer graphics, human-computer interaction, or database technologies.
You are expected to bring or develop the necessary soft skills for working in teams, as well as for managing and communicating your research progress. The same holds for the necessary hard skills in software development and scientific writing.

Place of Employment and Place of Work:
The place of employment is Aarhus University, and the place of work is Department of Computer Science, Åbogade 34, 8200 Aarhus N, Denmark.
Computer Science started at Aarhus University in 1968 as a part of the Department of Mathematical Sciences. In 1998, computer science became an independent department at Aarhus University (www.cs.au.dk). Today, the department has 125 employees with a great mix of nationalities, and 600 BSc and Msc students on the programs Computer Science and IT product development.
As a PhD student, you are a valuable part of the department. All PhD students get a support group of experienced advisors outside their own research group. The main purpose of the support group is to give feedback on the PhD work to help you reach your full potential as PhD student, and to ensure that any obstacles that may arise are overcome as smoothly as possible. Another important aspect is socializing with your peers across the department. The department host an annual retreat for PhD students and Postdocs, and social events are continuously organized by the Junior Club, which is run by PhD students.
The department is strong in both theoretical and experimental computer science. In recent years, we have seen close cooperation between different research groups – even those that traditionally are perceived as being far from each other. As we emphasize multidisciplinary attitudes to research, no firm dividing lines are drawn between the various strands of subjects, and there is a lively interaction between all research areas. Problem-oriented and inter-disciplinary approaches characterize our research.
Methoden:
Anfangsdatum: 1. Februar 2020
Geschätzte Dauer: 3 Jahre
Bezahlung: ca. 31.000 DKK/Monat brutto
Papers:
Sonstiges:

Archiv-Übersicht