PUB - Publikationen an der Universität Bielefeld

This project aims to explore the use of biologically-inspired spiking neural network (SNN) architectures in hardware for efficient and online learning using computer vision-based applications. The project is divided into two sub-projects, with one focusing on the design space exploration of possible computer vision applications and analyzing resource-efficient implementations of biologically-inspired SNNs, while the other focuses on testing SNN-specific online learning methods directly on hardware. The project will evaluate different neuron and synapse models and configurations of these models, coupled with event-based cameras to evaluate the solutions based on practical applications in a real-time environment. The project has achieved preliminary results, including the development of a Binary Neuronal Associative Memory (BiNAM) using SNNs and the creation of a novel SNN simulator tool called RAVSim, which provides a runtime environment to analyze and simulate the SNN model. The RAVSim tool is publicly available as an open-source simulator. The ultimate goal of the project is to deliver a system that can be used for ultra-high-speed computer vision applications using Spiking Neural Network-based Multi-Core Architecture.