Optical computation based on neuronal signal processing enhances energy efficiency
The dendrites of neurons form a highly branched tree-like structure largely devoted to receiving incoming signals from other neurons and processing them. In fact, some estimates suggest that approximately 75 % of the dendritic surface participates in synaptic transmission, or the reception of a signal from other neurons. The processing itself has linear and non-linear, analog and threshold features, as well as being complex and efficient. Models have been exploring the capabilities of such dendritic computation. Scientists working on the EU-funded ADOPD project intend to harness these concepts for ultrafast fibre-optical computational units, a basis for next-generation neuromorphic computing with significantly reduced energy consumption compared to standard computers. As end-users demand the ever faster processing of exponentially increasing amounts of information for applications ranging from cell phones and gaming to self-driving cars, novel computational units based on dendrite-inspired fibre-optical systems could be a game changer.
Fields of science
Call for proposal
See other projects for this call
Funding SchemeRIA - Research and Innovation action
96524 Neuhaus Schierschnitz