Studentische Hilfskraft (m/w/d): Evaluation of Automated Machine Learning Tools for Image Classification

  • Pa

Webseite FraunhoferIEM Fraunhofer-Institut für Entwurfstechnik Mechatronik IEM

Image classification is a well-established discipline in the machine learning community. We see the results of the corresponding research in our daily life, and everybody already has manually classified images to prove not to be a bot. However, the design of the models for image classification is a time-consuming task that requires expert knowledge. AutoML – and particularly neural architecture search – techniques aim at automating this task. There already exist several frameworks, tools, and services like AutoKeras or google autoML vision in this context. What is missing is a comparison of these tools with respect to their performance and their resource consumption.

 

Your tasks

  • Research exiting frameworks/tools for algorithm selection and hyperparameter optimization or neural architecture search for image classification
  • Compare the frameworks/tools empirically based on several datasets that will be provided
  • Construct an infrastructure that eases your experiments in terms of providing the necessary computing resources (e.g., based on Kubernetes) and logging your experiment setup and results (e.g., via mlflow)

Your Profile

  • You study computer science, business informatics or a comparable course of studies
  • You have programming experience in python
  • You have experience with ML-libraries like scikit-learn or keras
  • You are a fast adopter to new technologies
  • You are motivated and you like working self-dependently
  • You are fluent in spoken and written German or English

We offer

  • Possibility for attractive theses
  • Flexible working hours

Your application

Please send us the following information and note that incomplete applications may not be respected for further consideration:

  • Short(!) motivation letter stating your experience relevant for this position
  • Curriculum vitae
  • Transcript of records (vorläufiger Notenspiegel) showing the grades of your courses at your current university
  • Any certificates relevant for the position

Contact person
Dr. Matthias Meyer
Fraunhofer-Institut für Entwurfstechnik Mechatronik IEM
Zukunftsmeile 1, 33102 Paderborn
Telefon: +49 5251 5465-122

Um sich für diesen Job zu bewerben, besuche bitte www.iem.fraunhofer.de.