Neuromorphic Computing Models: Using Inhibitory and Excitatory Neurons for Magnetic Domain Pattern Recognition

Neuromorphic computing, which emulates the processes of the nervous system, is increasingly crucial in enhancing the energy and operational efficiency of artificial intelligence technologies. This significance stems from its ability to detect rapidly changing signals amidst noisy backgrounds. In this seminar, we delve into the physical mechanisms through which the brain encodes, decodes, and processes information, utilizing fundamental building blocks such as excitatory and inhibitory neurons. We explore spiking neural networks (SNNs) through the Python library Brian 2, demonstrating the successful replication of neuronal behavior for the identification of simple patterns. Furthermore, we propose novel approaches that leverage such neuronal properties in solid-state architectures. These are designed for use in neuromorphic cameras to optimize real-time data acquisition in the analysis of magnetic domains and their temporal evolution, with a particular emphasis on Kerr microscopy images.