Both factors, multilayer Feedforward Neural Networks (FFNs) and short-term synaptic plasticity (STP), are considered crucial in the transmission and processing of neural signals. In this study, a 10-layer FFN was constructed to study the impact of STP on neuronal activity propagation. Neurons within the same layer do not have direct connections; instead, neurons between adjacent layers are randomly connected with a specific probability. The findings indicate that synaptic plasticity can regulate the network’s firing rate, synchronization, and output firing rate gain. Notably, short-term facilitation (STF) allows the network to exhibit high-pass filtering, while short-term depression (STD) achieves low-pass filtering. Combining STF and STD synapses in the FFN broadens the range of input firing rates effectively transmitted by the network. Increasing the proportion of STD-dominated synapses enhances the gain of low firing rate signals but reduces the gain of high firing rate signals. Adjusting the mix of synapses enables the network to implement bandpass filtering and control firing rate gain. These results underscore the effectiveness of modulating short-term synaptic plasticity in regulating neural activity propagation in FFNs.

中文翻译：

---

通过短期突触可塑性调节前馈神经网络中的尖峰传播


多层前馈神经网络（FFN）和短期突触可塑性（STP）这两个因素被认为在神经信号的传输和处理中至关重要。在本研究中，构建了10层FFN来研究STP对神经元活动传播的影响。同一层内的神经元没有直接连接；相反，相邻层之间的神经元以特定概率随机连接。研究结果表明，突触可塑性可以调节网络的放电率、同步性和输出放电率增益。值得注意的是，短期促进（STF）允许网络表现出高通滤波，而短期抑制（STD）则实现低通滤波。在 FFN 中结合 STF 和 STD 突触拓宽了网络有效传输的输入放电率的范围。增加 STD 主导的突触比例会增强低放电率信号的增益，但会降低高放电率信号的增益。调整突触的混合使网络能够实现带通滤波并控制放电率增益。这些结果强调了调节短期突触可塑性在调节 FFN 神经活动传播方面的有效性。