In the human brain, the time-dependent properties of interconnected neurons allow for the processing of information in time and space, due to both memory and computation functions being mixed together. In man-made devices, the physical separation of computing units and memories make it impossible to mimic these complex neural systems.
Researchers have been trying to imitate the integration of neurons and synaptic connections, but as the brain possesses more synapses than neurons, synapse-like devices are necessary to achieve this objective.
In trying to reach this goal, Alibart et al. have fabricated a nanoparticle (NP) organic memory field-effect transistor, which mixes NPs and molecules to implement computation and memory into one device. By changing the size of the NPs and of the transistors, the group managed to tailor the dynamic behavior of the device in the frequency/time domain. Using the charge-storage capability of the NPs and the amplification factor of the organic transistor, it was possible to mimic the short-term plasticity of a biological synapse using a hybrid device.
No comments:
Post a Comment