Emulation of Neural Dynamics in Neuromorphic Circuits Based on Memristive Devices
The most impressive properties of the human brain are widely acknowledged as being perception and consciousness. While the underlying mechanisms are not yet understood, it is very likely that neural dynamics, in connection with the topology of neural networks, may play a decisive role. Neuromorphic systems offer an interesting approach to emulate and model these processes, as they allow the complexity of neural networks to be mapped onto energy-efficient and real-time capable systems. For this purpose, analogue electrical circuits that are oriented as closely as possible to biological networks are investigated. Electronic devices are particularly important for this purpose, as they make it possible to emulate the mechanisms that are important to the learning and memory processes that occur at the connections of neurons in form of synapses. In this context, it has been shown that nano-ionic mechanisms, in socalled memristive devices, allow the emulation of synaptic plasticity on a descriptive level within a single device. Memristive devices are passive, non-volatile components whose resistance value depends on the applied electrical potentials. In recent years, the important plasticity mechanisms of synaptic information-processing have been emulated using memristive devices. The importance of memristive devices in terms of emulating dynamic processes within novel bio-inspired computing schemes attract worldwide interest and is the subject of this thesis.