📄 中文摘要
脉冲神经网络（SNN）通过稀疏的神经元活动、递归连接和局部学习规则，模仿生物神经网络以解决复杂的认知任务。这些机制为神经形态计算提供了设计原则，旨在应对现代计算中的能耗挑战。然而，大多数混合信号神经形态设备依赖于半监督或无监督学习规则，这在监督学习任务中优化硬件的效果不佳。这种缺乏可扩展的片上学习解决方案限制了混合信号设备在可持续智能边缘系统中的潜力。为了解决这些挑战，提出了一种新颖的学习算法，旨在提高脉冲神经网络的训练效率和硬件适应性。
用于在线和硬件感知训练的反馈控制优化器针对脉冲神经网络
📄 English Summary
A feedback control optimizer for online and hardware-aware training of Spiking Neural Networks
This research presents a novel learning algorithm for Spiking Neural Networks (SNNs) that mimics biological neuronal networks, which solve complex cognitive tasks through sparse neuronal activity, recurrent connections, and local learning rules. These mechanisms serve as design principles in Neuromorphic computing, addressing the critical challenge of energy consumption in modern computing. However, most mixed-signal neuromorphic devices rely on semi-supervised or unsupervised learning rules, which are ineffective for optimizing hardware in supervised learning tasks. This lack of scalable on-chip learning solutions restricts the potential of mixed-signal devices to enable sustainable, intelligent edge systems. The proposed algorithm aims to enhance the training efficiency and hardware adaptability of SNNs.