Structural Plasticity in Spiking Neural Networks on Neuromorphic Hardware
Goal of my work is to introduce structural plasticity observed in biologic neural networks, in neuromorphic spiking neural networks.
A structurally plastic network will be implemented in a system of neuromorphic- and FPGA hardware, that I am currently designing.
In order to give the network the ability to adapt its topology in an activity-dependent fashion, an algorithm on the FPGA will emulate the process of dendritic spine growth- and retraction with the help of models that are derived from recordings in neuronal cell culture.
The found models will be verified with help of RAMPs EOSFET/EOSC transducer array, that acts as an interface to a biologic cell culture.
Finally, the systems ability to optimize network-topologies will be exploited to efficiently make use of the limited resources provided by our artificial neuron array-chips.