Physical reservoir computing

Reservoir lab applies reservoir computing to efficiently use the dynamics of physical media for computing. The most prominent media we exploit for physical reservior computing are photonic media and compliant robot bodies.



Former project: compliant robotics: AMARSi FP7 
AMARSi is a EU-funded research project in the Seventh Framework Programme. The project is a large scale integration project hosted in the category Information and Communication Technologies (ICT), unit E5: Cognitive Systems, Interaction and Robotics.
Take a look at our robots on our robotics page!
Brain-inspired computation
Human Brain Project
In short, the goal of the Human Brain Project is to build a completely new ICT infrastructure for future neuroscience, future medicine and future computing that will catalyse a global collaborative effort to understand the human brain and its diseases and ultimately to emulate its computational capabilities.
The current role of Reservoir lab in the Human brain project is to combine neural control with morphological computation and physical reservoir computing to facilitate locomotion control (neurorobotics).
Photonic reservoir computing
In the IAP-project PhotonicsATbe, we are studying and building reservoirs using photonic components. This is the second IAP-project with this name, financed by the Belgian Science Policy Office (BelSPO).
More information on the current IAP project can be found at The previous IAP-project is documented at
Machine learning
Brain Computer Interfaces
The applications of state-of-the-art machine learning techniques to EEG based Brain Computer Interfaces (BCIs). Our aim is to reduce or eliminate training time while achieving or improve upon state-of-the-art performance. We focus mainly on the P300 speller and Motion Imagery.
Deep learning
We apply deep learning to various application domains such as music classification, compression and sign language recognition and regularly compete (with very good results) in Kaggle competitions using deep learning approaches.
Machine learning approaches to physical system optimisation
We have applied an extension of backpropagation through time to physical media to optimise the dynamical behaviour of various physical systems (mechanical, photonic).