在具有不同局部性的脉冲神经网络学习方法方面的基准测试Spiking Neural Networks (SNNs) achieve performance comparable to Artificial Neural Networks (ANNs) in machine learning tasks, with processing done through spikes in an event-based mechanism that reduces energy consumption. However, training SNNs is challenging due to the non-differentiable spiking mechanism, and alternative learning methods with varying degrees of locality have been proposed. This research explores the training process similarities, the influence of explicit recurrence, and the performance of local learning methods under adversarial attacks.
Feb, 2024