COMET-Project Sebista (2019-2022)

Secure Big Stream Data Processing

Motivation

Artificial Intelligence (AI) applications of big companies such as Google and Facebook profit most from the trend of massive data collection. Besides the huge benefits of the newly enabled applications, data privacy issues must be considered. Similarly, in the industrial field the trend continues to collect more and more data about processes and products. However, (justifiable) data privacy concerns often immensely reduce the amount of available data and prohibit innovative digitalization projects. The development of a secure distributed learning system will give machine operators control over their data and prevent their data from being used for unintended purposes.

A particularly difficult challenge with distributed learning algorithms are heterogeneous data sources which are predominant in real-world applications. Such scenarios require the development of heterogeneous transfer learning methods which can be used in combination with distributed learning.

The efficient storage of massive data from distributed sources (internet of things (IoT), industrial internet...) and the processing of the underlying time-series data in real time (stream processing) are the basis for many applications and new business areas. In this project, application-specific methods for the integration of heterogeneous data sources will be developed based on existing big data technologies, whereby particularly the management of the data quality is taken into account. Real-time processing is ensured by the use and development of efficient distributed algorithms.

Expected Results

  • State-of-the-art deep-learning algorithms for secure and collaborative models will be developed and leveraged for improved industrial applications
  • Extension of our online transfer learning method (GOTL) to heterogeneous transfer learning problems and distributed learning
  • Novel methods and tools for continuous data quality monitoring in large IoT systems which utilize non-relation data models (NoSQL), as typically encountered, e.g., in time-series data bases

Funding Partner

This project is subsidized in the frame of COMET – Competence Centers for Excellent Technologies by BMVIT, BMDW, State of Upper Austria and its scientific partners. The COMET program is handled by FFG.

Contact

Lisa Ehrlinger

Ehrlinger Lisa

Researcher Data Analysis Systems
Phone: +43 50 343 836

Christian Lettner

Lettner Christian

Researcher Data Analysis Systems
Phone: +43 50 343 837

back