Neuromorphic Computing

For information processing, we utilize so-called spiking neural networks (SNN), the third-generation of neural networks. The data between the information-processing units is coded in the form of pulses, similar to the human brain. Pulse-coded artificial neural networks make energy-efficient information processing possible, particularly via neuromorphic hardware.

We develop algorithms and software for highly energy-efficient neuromorphic hardware.

These activities will include

• exploring the opportunities and limits of calculations with SNN-based calculations
• porting conventional network algorithms over to SNNs and neuromorphic hardware
• develop new learning methods for SNNs
• create a software framework for learning from SNNs
• design innovative robot experiments in closed loops with real and virtual robots, environments and SNN implementations from learning and decisions processes

Neuromorphic hardware