Energy-efficient, compressed AI algorithms

In times of sustainability, companies are striving for energy-efficient "green" production. The demand for resource-saving hardware for the automated monitoring of production processes, machines or for traffic or construction-site monitoring is constantly increasing. This is reflected, among other things, in the increasing sales of mobile and edge devices capable of machine learning

Fraunhofer IDMT is currently developing methods for acoustic monitoring and event detection which can be implemented on mobile devices with low energy consumption and which process analysis data reliably. The compressed algorithms work without transferring information to the cloud, which has a positive effect on the latency of the application. In addition, there are fewer data protection and data security issues when sensor information is processed directly on the device.

Research

 

Research project

DMD4Future

Digitized material and data value chains

 

Research project

SEC-Learn

Sensor Edge Cloud for Federated Learning

Research project

TRA-ICT

Trusted Ressource Aware ICT