Spikes

In a thesis evolutionary methods to determine architecture and parameters of spiking neural networks (SNNs) should be investigated and applied to complex tasks, e.g., temporal sequence detection and robotic planning. Potentially, we can identify important SNNs parameters, which are responsible for solving a task better than a conventional ANN. Also, the observed behavior (of a a robot) could be linked to neural correlates in the SNNs, which leads the way to the field of computational neuroethology. Potential topics to be addressed in a thesis (obviously not all of them in a single thesis) are given in the following:

• SNN Models
• Evolution of SNN architecture and model parameters
• Temporal Sequence Detection
• Robotic Planning
• Comparison of evolved SNNs and conventional ANNs
• Spike train analysis of evolved SNNs
• Analysis of robotic behavior changes induced by intentional changes of SNN parameters