Impact of differential privacy procedures in data protection on the quality of trained AI models.

Context

The goal of the Privilege project is to build a care repository to collect data donations for research and improvement of AI models. For this purpose, the impact of differential privacy techniques on the quality of trained AI models will be investigated. The goal is to ensure privacy in the analysis of data without losing its informative value. For this purpose, different procedures will be analyzed and evaluated with respect to the quality of AI models.

Task s

• Within the project, the following challenges await you:
• Comprehensive research on the state of the art and science
• Development of a concept for the use of differential privacy methods
• Implementation of the investigated methods and execution of experiments
• Analysis of the results and evaluation of the impact on the quality of trained AI models
• Establishment of a maintenance repository for the collection of data donations
• Documentation of the project to enable sustainable further development

We OFFER

• An innovative research environment and a great team
• Insights into exciting research projects and valuable practical experience
• Flexible working arrangements and work-life balance
• Motivated and competent supervision

We expect

• Degree in computer science, electrical engineering, mechatronics or similar studies
• Knowledge in the field of artificial intelligence and machine learning
• Experience in programming with Python
• Ideally knowledge in the field of data analysis and processing
• Ability to work independently and on your own responsibility
• Very good knowledge of German or English
• Motivation and commitment

Required documents

We look forward to receiving your PDF application to Jens Juhl, juhl∂fzi.de, with the following documents:

• Current grade transcript
• Curriculum vitae in tabular form etc.